Article Images (0)
×![]()
featured_play_listSave article to folder
X
Create new folder
Full Text
PDF
Table of Contents
Abstract
Introduction
Results
Discussion
Author Contributions
Declaration Of Interests
Materials Availability
Data And Code Availability
Experimental Model And Subject Details
Cell Culture
Method Details
Generating P21-MiniIAA7- MTurquoise2 HL-7702 Cell Lines
Generating At AFB2 HL-7702 (HL-7702 P21-ITA) Stable Cell Lines
Generating DHB-Cdt1(30-120)-H1 HL-7702 P21-ITA Stable Cell Lines
Generating 53BP1 DHB-Cdt1(30-120)-H1 HL-7702 P21-ITA Stable Cell Lines
Generating CDK2 And CDK4/6 Sensors HL-7702 P21-ITA Stable Cell Lines
Generating KRAS HL-7702 P21-ITA Stable Cell Lines
Chemical Regents
Time-Lapse Microscopy
Image Analysis
RNA Sequencing And Data Analysis
Western Blotting
Co-Immunoprecipitation
Flow Cytometry
System For Tracking And Depletion Of P21 During P21-Induced Quiescence Arrest States.
Abstract
p21 inhibits CDK2 activity to induce quiescence in response to stress or developmental stimulation. It is currently unclear whether p21 exhibits an equal functional importance across different stages and states of the quiescence arrest. Here employing a rapid p21 degradation system, we evaluate the contribution of p21 across heterogeneous quiescence arrest states during quiescence progression. Our findings reveal that cells exhibit a dynamics dependency on p21 during quiescence arrest. At low levels of p21, quiescence is exclusively dependent on p21-mediated inhibition of CDK2 activity to prevent cell cycle progression. In contrast, when p21 accumulates to higher levels, quiescence transitions into an “auto-maintenance” state where p21 becomes less essential. Mechanistically, we found an active attenuation of the KRAS/ERK signalling pathway as a driver of reduced proliferation potential in this “auto-maintenance” state. This attenuation reinforces the robustness of quiescence through a mechanism that is independent of p21. Our results thus support a dynamic, adaptive mechanism for quiescence regulation that synchronizes the anti- and pro-proliferation signals. This mechanism is applicable over various stress or developmental quiescence context, offering a basis for cells to explore distinct quiescence states to achieve different degrees of robustness in cell cycle arrest.
Introduction
In response to a wide range of intra- and extracellular signals such as cellular stress and developmental cues, cells encounter a pivotal decision point of either to proliferate or enter quiescence 1 , 2 . Cells enter quiescence to overcome stress or prepare for reaction to stimuli 3 – 6 . Unlike senescence, quiescent cells can be reverted back to proliferation when external stimuli are removed 7 . The primary mechanism triggering quiescent arrest is through halting the activation of cyclin D-CDK4/6 complexes and cyclin A/E-CDK2 complexes 8 . The lack of CDK activity restricts cell cycle entry through preventing RB phosphorylation and E2F1 activation 9 . Thus, either the absence of pro-proliferative factors or the expression of cell cycle inhibitors, such as p27 or p21, can drive cells into quiescence 10 , 11 . Once entering the quiescent arrest, further maintenance of this state is traditionally thought as being passively sustained by the quiescence initiation mechanism. However, in quiescence induced by the depletion of pro-proliferative factors, a long-term depletion is also found to inhibit lysosomal activity, leading to a deeper quiescence state that further stabilizes the arrest 12 . In the case of stress induced quiescence mediated by cell cycle inhibitors, the maintenance mechanism remains unexplored. A widely accepted hypothesis posits that cell cycle inhibitors are induced to counteract pro-proliferative signals, acting as a “barrier” to proliferation 13 . According to this hypothesis, the removal of these inhibitors is expected to permit quiescent cells to re-enter cell cycle. Among cell cycle inhibitors, damage- or stress-induced or developmental quiescent arrest is induced by a pivotal cyclin-dependent kinase inhibitor p21 14 – 16 . p21 is responsive to various environmental or cellular stresses 17 , and it has a crucial role for stem cell quiescence, including NSCs and HSCs 18 , 19 . p21 can be activated by p53, TGF-β/SMAD or PI3K/FOXO pathways 20 – 22 , to halt cell cycle progression through physically sequestering CDK2 and CDK4/6, and inhibiting their kinase activities. The intricate interplay between p21 and CDK activities was regarded as the key determinant for quiescence-proliferation decision 23 . Importantly, knockout studies show that p21 depletion alone is sufficient to abolish quiescent arrest 16 . Utilizing live-cell p21 reporters to track endogenous p21 levels, it is observed that even low levels of p21 protein can suppress CDK2 activity and effectively induce quiescence 14 , 15 . Recent studies suggest that p21 levels and its accumulation dynamics exhibit heterogeneity and are tightly associated with cell cycle arrest and recovery fate. For example, in a recent study using multiplexed single cell imaging, it was shown that high p21 levels can drive cells into a deeper arrest state toward senescence. Interestingly, these senescent cells exhibit a hallmark of having high levels of both p21 and pro-proliferative factors 11 . In another study evaluating the cancer cell chemotherapy, it was found that cells accumulate high- or low-p21 levels are driven into senescence, while those with intermediate levels of p21 can recover from the arrest upon subsequent drug withdrawal 24 . These lines of evidence suggested that the heterogeneity in p21 accumulation has a significant effect on the subsequent cell fate determination. However, the precise molecular underpinnings of arrest states induced by different levels of p21 during the time of its accumulation remain to be resolved. Here we seek to evaluate the following questions: 1) What is the dynamic pattern of p21 in heterogeneous quiescent arrest states induced by different stress and signals? 2) Does p21 carry equivalent functional weightage in these heterogeneous quiescent arrest states? 3) How does p21-mediated arrest engage other pathways under distinct arrest states? We address these questions using a quantitative single-cell live imaging approach in combination with a method of acute p21 depletion. We follow the recovery potentials of quiescent cells upon p21 depletion. Our analysis identifies cells with distinct recovery capabilities that are indicative of distinct arrest states associated with different p21 levels. At low levels of p21, quiescence is dependent on p21-mediated inhibition of CDK2 activity and its removal allows cells to re-enter proliferation. At high levels of p21, cells transit into an “auto-maintenance” state where p21 action is no longer essential and its removal becomes insufficient to allow cells to re-enter cell cycle. Importantly, these cells do not exhibit any of the previously reported hallmarks of senescence. Further analysis reveals a compromised RAS/ERK signaling pathway, reactivation of which rescues cells’ ability to re-enter proliferation upon p21 depletion. Thus, our approach of using a rapid p21 depletion method has unveiled important new insights about the dynamic requirement of p21 during quiescence arrest states. They underscore a critical role of crosstalk between anti- and pro-proliferative signaling in dictating the long-term maintenance of quiescent arrest states.
Results
Dynamics of p21 accumulation under different quiescence arrest conditions
The aim of this study was to evaluate the functional relationship between p21 accumulation and quiescence. We utilized a p21 reporter suitable for both live-imaging detection and an induced degradation 25 . Here a HL-7702 cell line was established that contained both alleles of CDKN1A , the p21-encoding gene, endogenously tagged with miniIAA7-mTurquoise2 ( Fig. S1A-C ; also see Methods for details). This cell line was also designed to contain a CDK2 activity indicator for defining quiescence-proliferation states 26 , as well as mMaroon1-tagged H2B and mKO2-tagged Cdt1 (aa30-120) as nuclear and G1/S transition markers 27 , respectively ( Fig. 1A , S1D, E ; also see Methods and legend for details). Several stress contexts are well known to cause p21 induction and accumulation ( Fig. 1B ; S1F ). Specifically, we used A-769662 to activate AMPK and mimic nutrient limitation ( Fig. S1G ), topoisomerase II inhibitor Doxorubicin, the DNA double-strand break inducer Zeocin, and Palbociclib to induce DNA damage 28 . We also used Nutlin-3 to directly activate p53 induced p21 expression. In addition, our collection included TGF-β to induce p21 in an p53-independent way by promoting the transformation into a quiescent mesenchymal state 29 – 32 ( Fig. S1H ). We used this collection of p21 inducing agents to obtain a wide spectrum of p21 and, importantly, to facilitate cross-validating our findings. We performed live cell imaging detecting CDK2 activity, Cdt1 and p21 for cells under different conditions ( Fig. 1C , S2A ). Under unperturbed condition, the proliferation percentage of both mother and daughter cells exceeded 80% and 70% respectively ( Fig. S2B, C ), with an average cell-cycle length of 28 hours ( Fig. S2D ). Under various stimuli to induce p21 activation, the cell cycle length and proliferation rate were elongated or compromised to varying degrees ( Fig. S2B-D ). Across all treatments, p21 accumulation was found to take place in G0/G1 phase ( Fig. 1D ), and it exhibited a mild positive correlation with the length of G0/G1 phase (Pearson correlation 0.2512, p<0.0001) ( Fig. S2E ). In contrast, proliferative cells had a minimum level of p21 during their G1 phase despite the heterogeneity in G1 length ( Fig. S2F, G ). These results documented an up-regulation of p21 levels and its association with cell cycle arrest induced by different pathway stimulation in our experimental setup. At the single cell level, p21 expression exhibited heterogeneous accumulation dynamics in cells that entered quiescence ( Fig. 1E ). p21 typically followed an increase-plateau accumulation kinetics in a majority of the cells analyzed, while other cells also exhibited a continued accumulation over the course of our observation. We found that both variations in the external stress level ( Fig. S3A-D ) and accumulation time ( Fig. S2E ) contributed to the heterogeneity in p21 levels during the quiescent arrest state. A concurrent monitoring of p21 and CDK2 dynamics revealed that p21 accumulation and CDK2 activation exhibited a mutually exclusive relationship ( Fig. 1F ), confirming that a low level of p21 is sufficient to inhibit CDK2 activity in our system as reported previously 15 .
Heterogeneous p21 levels associated with distinct quiescent arrest states
To investigate whether cells with different p21 levels exhibit differential molecular features towards the quiescence arrest, we induced p21 with Palbociclib or Nutlin-3, and sorted cells based on p21 fluorescence levels to obtain p21-negative, p21-low, and p21-high sub-groups ( Fig. 2A ). p21 has been reported to inhibit several cyclin-dependent kinases (CDKs), including CDK4/6 33 , CDK2 26 , and CDK1 34 , through competitive binding with cyclins. We firstly performed an immuno-precipitation analysis to evaluate the interaction between p21 and different CDKs in cells from these sub-groups. Our results show that p21 in cells from both the p21-low and p21-high sub-groups interacted with CDK2 ( Fig. 2B, C ). However, only in the p21-high sub-group did p21 interact with CDK4 and CDK6 ( Fig. 2B, C ). To follow the dynamics of CDK4/6 activity in relation to p21, we further integrated mKO2-tagged CDK4/6 sensors in HL-7702 p21-ITA cells with mVenus-tagged CDK2 sensors ( Fig. S4A ). Time-lapse imaging showed that the CDK4/6 sensor activity increased prior to the increment of CDK2 in untreated control cells ( Fig. S4B ), consistent with previous studies 35 . Upon p21 induction with Nutlin-3 treatment, cells that had a prominent decrease in CDK4/6 activity exhibited a high p21 level ( Fig. S4C ), supportive of a differential inhibitory effect of p21 on different CDK activities. To evaluate whether variations in p21 accumulation are associated with varied cellular states at the transcriptomic level, we performed an RNA-seq analysis of quiescent cells with different p21 levels ( Fig. 2D ). This analysis identified genes in clusters 2 and 3 whose expression positively correlated with p21 level, whereas those in clusters 1 and 4 had a negatively correlation with p21 levels ( Fig. 2D ). Cluster 2 and 3 genes featured p53 pathway genes, as well as lysosome pathway genes which had previously been reported to be associated with a deeper quiescence arrest state induced by growth factor depletion. Genes in clusters 1 and 4 included proliferation and cell cycle-related genes, suggestive of a stronger inhibition of cell cycle in p21-high cells ( Fig. 2D ). Importantly, the cellular senescence pathway genes were down-regulated in p21-high cells, suggesting that these cells were channeled away from senescence at the initial stage of quiescent arrest. Indeed, senescence markers such as SA-β Gal did not increase in the p21 expressing cells relative to the control group ( Fig. 2E ).
Differential p21-dependency during the quiescence arrest unveiled from p21 depletion analysis
So far, our data suggested heterogeneous cell states associated with different p21 levels during quiescence arrest. We performed experiments to further investigate whether p21 might carry an equal functional weightage in these heterogeneous quiescent arrest states. Here we employed the p21 knock-in reporter containing the miniIAA7-mTurquoise2 for induced and rapid degradation of p21 ( Fig. S5A ). In our analysis, the endogenous p21 protein became undetectable within one hour after the addition of IAA ( Fig. S5B ), and this degradation was dependent on the ubiquitin pathway as expected (see Fig. 3A, B and legend for details). At the population level, degradation of p21 under different treatments reduced the percentages of cells in the G0/G1 phase, highlighting the role of p21 expression in influencing the quiescent-proliferative cell fate choice ( Fig. S5C ). We followed p21 depletion in live imaging experiments. Cells treated with A-769662 for 48 hr without IAA addition remained in the quiescent state ( Fig. S6A ). Upon IAA addition to deplete p21, cells exhibited varied responses in cell cycle progression as indicated by CDK2 accumulation kinetics. While a fraction of cells exited quiescence and re-accumulated CDK2 activity, others remained arrested. These latter cells exhibited either a shallow CDK2 activity increase followed by a reduction to background levels, or a complete lack of recovery in CDK2 activity ( Fig. 3C, D ; Fig. S6B ). We refer to these three types of CDK2 recovery kinetics upon p21 depletion as “recovery”, “abortive recovery” and “arrest”, respectively. The observation of these three distinct types of CDK2 recovery kinetics supported the existence of heterogeneous arrest mechanism among the quiescent cells prior to p21 depletion. These three types of recovery kinetics were similarly observed across all the indicated treatments ( Fig. 3E ). Notably, cells in the “abortive recovery” sub-group accounted for more than 50% of the cells, consistently representing a higher percentage than the other two sub-groups ( Fig. 3E ). The detection of cells in the “abortive recovery” and “arrest” sub-groups further suggested that, these cells, unlike those in the “recovery” sub-group, had been in a quiescence state(s) that was no longer reversible upon p21 depletion.
Different p21-dependency in quiescent cells is associated with distinct levels of p21 at the time of its depletion
To better evaluate p21-mediated quiescent arrest states, we compared specifically the behavior of cells exhibiting the three distinct kinetics of CDK2 activity upon p21 depletion. These three groups of cells had different overall p21 levels prior to IAA treatment with the “recovery”, “abortive recovery” and “arrest” sub-groups having the lowest (<40 a.u.), intermediate (∼40-60 a.u.), and highest (>80 a.u.) overall levels of p21, respectively ( Fig. 4A ). This relationship was observed consistently in all conditions inducing quiescent arrest in our analysis. To determine whether differential recovery outcomes upon p21 depletion manifest different ranges of existing p21 levels, IAA was added at different time points after stress treatment, allowing cells to accumulate p21 to different levels prior to depletion ( Fig. 4B ). In particular, IAA addition at an earlier time point (1 hr) corresponded with a much lower existent p21 level in the population than such a treatment at a later time point (12 hr) ( Fig. 4C ; Fig. S7A ). Despite this difference, we still observed the same three sub-groups of CDK2 recovery kinetics, only with a shift in the proportions of individual sub-groups ( Fig. 4D ; Fig. S7B ). In particular, the early p21-depletion treatment led to a larger proportion of the “recovery” sub-group accompanied by a smaller proportion of the “abortive recovery” sub-group ( Fig. 4D ; Fig. S7B ). Despite a lower amount of p21 across the population under the early p21-depletion treatment, the measured levels of p21 associated with the three sub-groups of CDK2 recovery kinetics remained largely unchanged ( Fig. 4E ; Fig. S7C ). These results are supportive of the idea that varied quiescent arrest states as measured by the CDK2 recovery kinetics was associated with cells’ existent p21 level at the time of its depletion.
Cells with high levels of p21 exhibit attenuated ERK activity
Our experiments described above suggested a differential inhibition to CDK4/6 at high levels of p21 ( Fig. 2B, C ; Fig. S4C ). To investigate whether such an inhibitory effect may contribute to a failure to recover upon p21 depletion in the p21-high group, we tracked the dynamics of CDK4/6 in the three sub-groups of cells after p21 degradation. Upon p21 removal, only the “recovery” sub-group exhibited a prominent increase in CDK4/6 activity ( Fig. 5A ). CDK4/6 is known to promote cell cycle entry by phosphorylating RB to bypass the restriction point 36 . We evaluated phospho-RB and cyclin E1 status under our experimental setup, and found that pRB exhibited a significant increase after p21 depletion only in cells from the p21-low group ( Fig. 5B ). In the p21-high group, pRB not only failed to exhibit an increase upon p21 removal but instead, showed a further decrease ( Fig. 5B ). These results were indicative of a loss of capacity to reactivate CDK4/6 in the p21-high group even after p21 was depleted, consistent with the loss of proliferation potential of cells in the “abortive recovery” and “arrest” sub-groups. In our RNA-seq data described above we also observed a reduction in the expression of pro-proliferative genes in p21-high cells, including KRAS, MYC ( Fig. 2D ). This suggested that cells with high p21 levels may have a reduced pro-proliferation signalling pathway activity. Both KRAS and MYC were part of the ERK signalling pathway, one of the most important pro-proliferation signalling pathways. We thus checked the phospho-ERK status, which serves as an indicator of ERK pathway activity 37 ( Fig. S8A ). We found that phospho-ERK was dramatically down-regulated in p21-high cells ( Fig. 5C ), even after p21 degradation ( Fig. 5D ). These results suggested that RAS/ERK attenuation in p21-high group of cells contributed to the inability of recovery even after p21 removal. To evaluate whether RAS/ERK signaling attenuation is functionally relevant to the quiescent state of p21-high cells, we engineered a dox-inducible G12V mutant KRAS with a BFP-tagged cell line suitable for reactivation of the RAS/ERK pathway after p21 depletion 38 , 39 . The G12V mutant provides constitutive KRAS activity independent of external growth factor sensing. In our system, KRAS (G12V) could be induced in less than 8 hours with doxycycline treatment, as observed in both time-lapse imaging and western blot ( Fig. 5E , S8B ). KRAS (G12V) expression upon p21 depletion under Palbociclib treatment reduced the percentage of the “arrest” and “abortive recovery” sub-groups while increasing the percentage of the “recovery” sub-group to more than 50% ( Fig. 5F ). With increasing doxycycline concentration, the percentage of the “recovery” sub-group further increased from ∼ 50% to ∼ 65% ( Fig. S8C, D ). The ability of KRAS (G12V) to rescue CDK2 recovery kinetics was also observed under Nutlin-3 treatment condition ( Fig. S8 E-G ). These results demonstrated that reactivation of the RAS/ERK pathway could rescue the persistent arrest and revive CDK2 activity. In a complementary experiment, we inhibited ERK using Ulixertinib, an ERK1/2 inhibitor which can induce G1 cell cycle arrest ( Fig. S9A ), at the time of p21 depletion. We observed a marginal effect on the “recovery” sub-group, but a slight decrease in “abortive recovery” sub-group ( Fig. 5G ). In contrast, when inhibiting ERK at the beginning of Palbociclib treatment, the CDK2 recovery rate upon p21 removal was severely affected ( Fig. 5H ). Over 80% of cells exhibited sustained low levels of CDK2 and CDK4/6 activities, and importantly, such induced deficit could not be rescued even with a subsequent expression of KRAS (G12V) ( Fig. S9B ). These results suggested that an appropriate RAS/ERK pathway activity at the time of p21 depletion was important to capacitate the recovery upon p21 depletion.
A reduced sensitivity to grow factors contributes to the “auto-maintenance” state in p21-high cells
The restoration of ERK activity and the recovery potential observed with KRAS (G12V) above indicate that, the downstream MEK-ERK signalling pathway remains functional and capable of responding to activated RAS. On another hand, p21-high cells, which exhibit reduced ERK activity, showed no change in external growth factor concentration. This implies that these cells may have become desensitized to pro-proliferation signals due to impaired growth factor sensing required to activate RAS ( Fig. 2D ; 5C ). Such a desensitization could shift the cells into an “auto-maintenance” state, an arrest state that is no longer dictated solely by p21. To test this possibility, we investigated whether elevated external growth factor levels could counteract the attenuated ERK signalling in p21-high cells. Here we evaluated the effect of raising serum levels towards promoting the recovery following p21 depletion ( Fig. S10A ). Raising serum concentration from 0% to 20% indeed increased the expression of phospho-CDK2, phospho-RB, and cyclin D1 over time ( Fig. S10B ). After Nutlin-3 treatment for 48 hours, depleting p21 with IAA in conjunction with an increase of serum from 10% to 20% promoted the expression of phospho-CDK2, phospho-RB, and cyclin D1, as well as mitotic marker cyclin B1 ( Fig. 6A ). Next, we tracked the CDK2 activity in live imaging at the single cell level. 20% of serum was added at different time points after p21 depletion ( Fig. 6B ). CDK2 activity in both the “abortive recovery” and “arrest” sub-groups increased upon increasing serum to 20% at either time point ( Fig. 6C, D ; Fig. S10C-F ). The percentages of high serum-induced recovery in the “abortive recovery” sub-group were consistently higher than those in the “arrest” sub-group ( Fig. 6C, D ; Fig. S10C-F ). Additionally, the recovery percentages had a greater increase when serum concentration increase was at an earlier time than at a later time ( Fig. 6C, D ; Fig. S10C-F ). Collectively, these results suggest that a reduced sensitivity to pro-proliferation signals in p21-high cells pushes these cells into an “auto-maintenance” arrest state that is no longer dictated solely by p21. They illustrate a synchronized regulation between the anti- and pro-proliferation machineries in reinforcing the quiescence arrest.
Discussion
Quiescent arrest is a temporary exit from the proliferative cell cycle, traditionally considered to employ universal arrest mechanism 23 . However, several recent studies point toward the existence of more stably quiescent arrest states in tumor 40 or stem cells 41 , 42 . Furthermore, even though low levels of p21 is sufficient to induce quiescent arrest by selectively binding to CDK2 and competing with cyclin A/E 14 , recent single-cell studies have highlighted a relationship between early-stage p21 heterogeneity and the potential to proceed to senescence under non-lethal chemotherapy conditions 24 . Thus, it is currently unclear whether heterogeneity in p21 accumulation may elicit different quiescent arrest states. Our analysis of the recovery potentials upon p21 depletion suggests that p21-mediated quiescence has distinct arrest states that elicit different affinities for CDK inhibition and display distinct transcriptomic profiles. More importantly, quiescent cells in heterogeneous arrest states are under the control of different anti-proliferation mechanisms with varied stability of quiescence arrest. One hypothesis regarding p21 heterogeneity during quiescence is that high levels of p21 may induce senescence rather than quiescence, preventing CDK2 activity recovery even after p21 removal. In our experiments, p21-high cells have an up-regulation of lysosome related genes, which have also been previously observed in deep quiescence induced by serum depletion, a quiescent state leaning toward senescence 12 . However, there is no evidence of up-regulation of the senescence marker SA-β Gal in our system ( Fig. 2E ). In fact, senescence-related genes are down-regulated in our RNA-seq data ( Fig. 2D ). Moreover, single-cell experiments with serum stimulation and Tet-on KRAS (G12V) expression show that “abortive recovery” and “arrest” sub-groups of cells retain a strong ability to re-enter the cell cycle under appropriate conditions. The role of p21 during quiescent arrest is generally regarded as a direct inhibitor of proliferation and, consequently, its removal is expected to lead to cell cycle reactivation 43 . However, using our AID system that affords a rapid degradation of endogenous p21, we observed diverse responses in cell cycle progression upon p21 depletion. Notably, cells exhibiting different recovery kinetics have correspondingly different overall p21 levels prior to depletion. We found that cells in the “abortive recovery” and “arrest” sub-groups remained in arrest even after p21 depletion, and this is because their pro-proliferation ERK signaling pathway remained low even after p21 removal. Collectively, our results suggest that in cells with modest to high levels of p21, a reduced ERK activity can further restrict CDK2 and CDK4/6 reactivation. This effect represents an additional layer of proliferation-restriction that synchronizes with p21 but is beyond the direct inhibitory effect of p21 ( Fig. 6E ). KRAS, a proto-oncogene and small GTPase, plays a crucial role in regulating signaling pathways that determine the quiescence-proliferation decision 38 . External mitogens activate the RAS/ERK pathway, promoting cell cycle entry. Elevated KRAS activity can drive quiescent cells into S phase, while KRAS depletion prevents this transition 44 . Interestingly, in anti-tumor therapies, inhibiting CDK2 alone leads to a transient suppression of CDK2 activity, only to be followed by a rebound within hours of the treatment due to the residual CDK4/6 activity maintained by the RAS/ERK pathway 45 . In contrast, the quiescence maintenance mechanism employed by p21-high cells utilizes a desensitized growth factor response to reduce RAS/ERK activity, an additional layer of deficit that goes beyond p21’s inhibition on CDK2. This two-tiered regulation provides a more robust quiescence arrest when necessary, thus preventing undesired pro-proliferation signals from overriding the arrest. In summary, our study underscores p21 heterogeneity in a two-tiered regulatory mechanism during quiescence arrest. While p21 has been previously shown to be closely associated with quiescent cell fate with a significant heterogeneity during the arrest, its functional impact across varied quiescence arrest states has remained largely unexplored. Our study, which combines precise time-lapse imaging and the AID rapid protein degradation system, provides new insights into a dynamic dependency on p21 during quiescent arrest. This understanding may also open up new strategies for enhancing the effectiveness of anti-proliferation treatments.
Author contributions
D.Z., F.H., H.L., and J.M. conceived the study and designed the experiments; D.Z., S.Q., Y.S., Z.A., C.M., and W.M. performed experiments and generated data; D.Z. analyzed the data and generated all figures; H.L. and J.M. acquired funding; D.Z., F.H., H.L., and J.M. wrote the paper and all approved the paper.
Declaration of interests
The authors declare that they have no conflict of interest.
Materials availability
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Hongqing Liang ( lianghongqing@zju.edu.sg ).
Data and code availability
Data: The authors declare that all data supporting the findings of this study are available within the paper and its supplementary information. RNA sequence raw data is deposited at SRA ( http://www.ncbi.nlm.nih.gov/sra ) with the BioProject ID: PRJNA1174361. Data are also available upon request. Code: RNA-seq analyses were performed using standard software packages. Code for live imaging data analysis is available upon request.
Experimental model and subject details
Cell culture
HEK293T and HL-7702 cells were maintained in DMEM (11965092; Thermofisher) supplemented with 10% FBS (SE100-B; VisTech), penicillin/streptomycin (100 U/ml each) at 5% CO2 and 37°C. Media were changed every 2 days and passaged every 2– 3 days. To induce epithelial-mesenchymal transition (EMT), HL-0077 cells were treated with 10 μg/ml TGF-β, with media changes every other day.
Method details
Generating p21-miniIAA7- mTurquoise2 HL-7702 cell lines
The generation of endogenously tagged p21-miniIAA7-mTurquoise2 cells was as reported in our previous studies 25 . In brief, the endogenous CDKN1A gene was tagged at the C terminus using CRISPR-mediated gene tagging. The gRNA for CDKN1A nearby the stop codon was used. The p21-gRNA plasmid (5’-GGAAGCCCTAATCCGCCCAC-3’) and the linearized PCR product of p21-miniIAA7-mTurquoise2-IRES-pruoR donor sequence was transfected into HL-7702 cells at the ratio of 1:1 using jetPRIME, according to the manufacturer’s instructions (101000046; Polyplus). Cells were treated with 10 μg/ml puromycin until outgrown of the colonies. Finally, homozygous knock-in clones were verified by PCR and western blotting.
Generating At AFB2 HL-7702 (HL-7702 p21-ITA) stable cell lines
HEK293T cells at approximately 70% confluence were transfected with psPAX2, pMD2.G, FUGW- At AFB2-mCherry-weak NLS at ratio 0.75:0.5:1 using jetPRIME (101000046; Polyplus), according to the manufacturer’s instructions. Two days after transfection, supernatant was filtered through a 0.45 μm polyethersulfone filter and was used to infect p21-miniIAA7-mTurquoise2 HL-7702 cells. After 7 days, mCherry-positive cells were sorted out via flow cytometry.
Generating DHB-Cdt1(30-120)-H1 HL-7702 p21-ITA stable cell lines
HEK293T cells at approximately 70% confluence were transfected with psPAX2, pMD2.G, and FUGW-DHB-mVenus or FUGW-H1-mMaroon1-IRES-mKO2-Cdt1(30-120) at the ratio of 0.75:0.5:1 using jetPRIME (101000046; Polyplus), according to the manufacturer’s instructions. Two days after transfection, supernatant was filtered through a 0.45 μm polyethersulfone filter and was used to infect HL-7702 p21-ITA cells. After 7 days, mVenus- and mMaroon1-positive cells were sorted out via flow cytometry.
Generating 53BP1 DHB-Cdt1(30-120)-H1 HL-7702 p21-ITA stable cell lines
HEK293T cells at approximately 70% confluence were transfected with psPAX2, pMD2.G, and FUGW-BFP-truncated 53BP1 (1220-1771) at ratio 0.75:0.5:1 using jetPRIME (101000046; Polyplus), according to the manufacturer’s instructions. Two days after transfection, supernatant was filtered through a 0.45 μm polyethersulfone filter and was used to infect DHB-Cdt1(30-120)-H1 HL-7702 p21-ITA cells. After 7 days, BFP-positive cells were sorted out via flow cytometry.
Generating CDK2 and CDK4/6 sensors HL-7702 p21-ITA stable cell lines
HEK293T cells at approximately 70% confluence were transfected with psPAX2, pMD2.G, and FUGW-DHB-mVenus or FUGW-H1-mMaroon1-IRES-mKO2-RB at the ratio of 0.75:0.5:1 using jetPRIME (101000046; Polyplus), according to the manufacturer’s instructions. Two days after transfection, supernatant was filtered through a 0.45 μm polyethersulfone filter and was used to infect HL-7702 p21-ITA cells. After 7 days, mVenus-and mMaroon1-positive cells were sorted out via flow cytometry.
Generating KRAS HL-7702 p21-ITA stable cell lines
HEK293T cells at approximately 70% confluence were transfected with psPAX2, pMD2.G, and PCW 57.1-BFP-KRAS(G12V) at ratio 0.75:0.5:1 using jetPRIME (101000046; Polyplus), according to the manufacturer’s instructions. Two days after transfection, supernatant was filtered through a 0.45 μm polyethersulfone filter and was used to infect DHB-Cdt1(30-120)-H1 HL-7702 p21-ITA or CDK2 and CDK4/6 sensors HL-7702 p21-ITA cells. After 7 days, cells were treated with 1 μg/ml Doxycycline for 24 hours and then BFP-positive cells were sorted out via flow cytometry.
Chemical regents
Drugs used in this study include: TGF-β (HY-P7118; MedChemExpress), A-769662 (S2697; Selleck), Palbociclib (S4482; Selleck), Nutlin-3 (S1061; Selleck), Doxorubicin (S1208; Selleck), Zeocin (R25001; Thermofisher), auxin (I5148; Sigma), Doxycycline (ST039A; Beyotime) and Ulixertinib (S7854; Selleck). TGF-β, auxin, and doxycycline were reconstituted in water, while the others were reconstituted in DMSO. All drugs were stored in aliquots at −20°C.
Time-lapse microscopy
A 24-well dish (142485; ThermoFisher) was pre-treated with 0.1% gelatin (#07903; StemCell). 2×10 4 cells were plated 24 hours before imaging in the 24-well dish. Cells were imaged in a humidified, 37°C chamber with 5% CO2. Images were taken every 15 minutes in the cyan (excitation CWL/BW: 440/10 nm; emission CWL/BW: 460/30 nm), green (490/10 nm; 535/30 nm), red (545/10 nm; 575/25 nm), and Cy5 (610/10 nm; 670/65 nm) channels using a Nikon Ti2-E inverted microscope (Nikon). The total light exposure time was kept under 500 ms for every snapshot, and the laser intensity for each channel was kept below 5%. To quantify p21 levels, the exposure time and intensity of the cyan channel were consistently set to 200 ms and 5%, respectively, throughout all experiments.
Image analysis
All image analyses utilized the Histone H1.0-mMaroon1 signal as a nuclear mask. We employed the automated segmentation software, Cellpose 2.0 46 , to segment individual cells within an image. For tracking single cells over time, we applied a mathematical framework known as the Linear Assignment Problem (LAP) 47 . CDK2 activity and CDK4/6 activity was calculated as the ratio between cytoplasmic and nuclear mean signals. The nuclear signal was measured within the representative region masked by H1.0-mMaroon1. The cytoplasmic signal was measured as a ring with an external diameter of 2 pixels from the nuclear boundary. CDK2 activity was used to distinguish cells in either quiescent or proliferative state. Quiescence was defined when the mean CDK2 activity stayed below 0.8 for more than 10 hours 26 . For p21 level quantification, the mean reporter intensity within the nuclear region was measured. The 53BP1 levels were quantified by segmenting foci using watershed algorithm and counting the number of foci per nucleus.
RNA sequencing and data analysis
Before library construction, a pellet of ∼100,000 cells were snap-frozen and stored at −80 °C. Total RNA was extracted from each sample using RNAiso Plus kit (9109; Takara). The cDNA was used for library construction using VAHTS Universal V8 RNA-seq Library Prep Kit for Illumina (NR605-01; Vazyme). The libraries were sequenced on Illumina NextSeq sequencer with PE150 chemistry. Every condition had two biological replicates. Overall sequence quality was examined using FastQC v0.11.2. Adaptor sequences were trimmed using Trimmomatic v0.32 48 . Reads were mapped to the human genome (GRCh38) using HISAT2 v2.0.3-beta 49 . Mapped fragments were counted using featureCounts 50 . Differential expression analysis was carried out using DESeq2 v1.10.1 51 . Genes with an adjusted p-value less than 0.05 and log 2 fold change more than 1 were considered to be differentially expressed. KEGG enrichment and heatmap analysis was carried out using ClusterGVis 52 .
Western blotting
Cells were collected and then directly lysed by addition of RIPA buffer (FD009; FUDE). Whole cell lysates were loaded onto 12% SDS-PAGE (FD346; FUDE) followed by transfer to PVDF membranes (IPVH00010; Millipore). After protein transfer, membranes were incubated in 5% milk in TBST at room temperature (RT) with rocking for at least 2 h. Primary antibody diluted in TBST was added and membranes were incubated overnight at 4°C with rocking. Membranes were washed for 5 min in TBST four times and anti-mouse (FDM007; FUDE) or anti-rabbit HRP-conjugated secondary antibodies (FDR007; FUDE) were diluted 1:10000 in TBST and incubated with membranes at room temperature with rocking for 1 hr. Membranes were washed for 5 min in TBST four times and visualized using ECL Western Blotting Substrate (E412-01; Vazyme Biotech). Blots were visualized using a gel imaging system (Clinx ChemiScope 3000 Series) and the image was analyzed by Gel-Pro analyzer software (v4.0.0.001). Antibodies used for include p21 (CST; #2947S), GAPDH (A19056; Abclonal), mCherry (26765; Proteintech), phophos-AMPK (#2535S; CST), AMPK (66536; Proteintech), CDK6 (14052, Proteintech), CDK4 (66950; Proteintech), p53 (A195851; Abclonal), E-cadherin (20874; Proteintech), N-cadherin (22018; Proteintech), Vimentin (A19607; Abclonal), phophos-CDK2 (AP0325; Abclonal), CDK2 (A16885; Abclonal), phospho-Rb (Ser807/811) (#2524S; CST), cyclin D1 (A19038; Abclonal), cyclin B1 (A22435; Abclonal), cyclin E1 (11554; Proteintech) and phospho-ERK1 (T202/Y204)/ERK2 (T185/Y187) (MAB1018; R & D).
Co-immunoprecipitation
Immunoprecipitations of p21, CDK2, and CDK6 from HL-7704 cells were performed using antibodies specific to p21 (CST; #2947S), CDK2 (A16885; Abclonal), or CDK6 (14052; Proteintech). The antibodies were captured using protein A/G magnetic beads (B23202; Selleck) according to the manufacturer’s instructions. Briefly, cells were collected and lysed directly in RIPA buffer (FD011; FUDE) on ice for 1 hour. The lysates were centrifuged at 4 °C for 10 minutes to remove cellular debris. A total of 500 μg of the cleared protein lysate was incubated with 5 μg of either rabbit IgG or the indicated antibodies at 4 °C overnight with rotation. The lysate was then incubated with 25 μL of protein A/G magnetic beads for more than 2 hours at room temperature, also with rotation. After incubation, the beads were washed with lysis buffer, and the bound proteins were eluted. The eluted samples were separated by SDS-PAGE, transferred to PVDF membranes, and probed with antibodies against p21, CDK2, CDK6, and CDK4. The antibodies used for Western blotting included p21 (CST; #2947S), GAPDH (A19056; Abclonal), CDK6 (14052; Proteintech), CDK4 (66950; Proteintech), and CDK2 (A16885; Abclonal).
Flow cytometry
Cells were fixed with 75% ice-cold alcohol at 4°C over-night and washed three times with ice-cold PBS. Then, cells were incubated with 50 mg/ml RNase A and 40 mg/ml propidium iodide (PI) (BB-4104; Bestbio) for 30 min at room temperature. Cell cycle analysis was done by Beckman Coulter DxFLEX Cell Analyser, and the further analysis and visualization were performed in Flowjo v10.8.1 software.
System for tracking and depletion of p21 during p21-induced quiescence arrest states.
Supplementary Video 1: live-cell imaging of Nutlin-3 treated p21-ITA HL-7706 cells with p21 depletion. The p21-ITA HL-7706 cell line, expressing Histone H1.0-mMarron1 (purple), a CDK2 sensor (green), Cdt1 (red), and p21-mTurquoise2 (blue), was treated with 10 μM Nutlin-3 for 24 hours with IAA added at the 12-hour. Live cell imaging was performed over 24 hours with images acquired at 15-minute intervals using a 20x objective lens. Supplementary Video 2: live-cell imaging of the 53BP1 sensor in p21-ITA HL-7706 cells under normal and Zeocin treatment conditions. The p21-ITA HL-7706 cell line, expressing Histone H1.0-mMarron1 (purple), CDK2 sensor (green), Cdt1 (red), p21-mTurquoise2 (blue), and 53BP1 sensor (yellow), was imaged over 48 hours with or without 50 μg/ml Zeocin treatment. Images were captured every 15 minutes using a 20x objective lens.
Article Images (0)
×![]()
Bioz is the world’s most comprehensive AI search engine for scientific experimentation. The Bioz search engine offers researchers billions of data-driven product, technique, and protocol recommendations. Bioz taps into the latest advances in Artificial Intelligence (AI) – including Natural Language Processing (NLP), Machine Learning (ML) and Generative AI to mine and structure hundreds of millions of pages of complex and unstructured scientific papers. By harnessing the power of AI, Bioz provides researchers with an unprecedented amount of summarized scientific experimentation knowledge right at their fingertips, ultimately helping to speed up drug discovery and the rate of success in finding cures for diseases. Bioz is used by over 15 Million researchers from over 17,000 different universities and companies in 196 countries.
