Irradiation-Induced Polyploidy Giant Cancer Cells Mediate Tumor Cell Repopulation via Neosis


 Background Tumor repopulation generally describes the phenomenon that residual tumor cells surviving therapies tenaciously proliferate and reestablish the tumor, presenting an embarrassing plight for cancer treatment. However, the cellular and molecular mechanisms underlying this process remains poorly understood. In this study, we proposed polyploidy giant cancer cells (PGCCs)-mediated and neosis-based tumor repopulation after radiotherapy.Methods The formation of PGCCs after irradiation was examined in vitro and in vivo. The demise of X-ray irradiated cells was detected by flow cytometry, clonogenic cell survival assay and transmission electron microscopy. Western blot was used to test cell proliferation and death related protein expression level of these irradiated cells. Time lapse microscopy was adopted to observe the destiny of PGCCs. The property of these PGCCs was identified by TUNEL assay, Brdu chasing assay, western blot, immunocytochemical and immunofluorescence staining. The relationship of HMGB1 with PGCCs-derived tumor repopulation was conducted via HMGB1 chemical inhibitors. Finally, animal model was used to verify the formation of PGCCs, and the relevance of HMGB1 in this process was investigated by immunohistochemical staining.Results The majority of PGCCs induced by irradiation move towards cell demise, whereas some of them intriguingly possessed proliferative property. Utilizing time-lapse microscopy and single-cell cloning assay, we observed that neosis derived from those PGCCs with proliferative capacity contributed to tumor cell repopulation after irradiation. Using the conditioned media collected from dying tumor cells to perform single-cell cloning assay, we unexpectedly demonstrated that HMGB1 released from dying tumor cells participated the process of neosis-based tumor repopulation. In irradiation treated animal tumor bearing model, the expression level of HMGB1 increased after irradiation compare with non-irradiated group. Moreover, some PGCCs presented high HMGB1 expression. Interestingly, we also observed that the proliferation potential of PGCCs varied. Some PGCCs proliferated at early stage, while some PGCCs proliferated at late stage.Conclusion X-ray irradiation could induce the formation of PGCCs, which could move towards both cell death and survival; irradiation-generating PGCCs mediated tumor cell repopulation after irradiation via neosis; HMGB1 released from dying cells stimulated the process of neosis and participated in tumor repopulation after irradiation.


Results
The majority of PGCCs induced by irradiation move towards cell demise, whereas some of them intriguingly possessed proliferative property. Utilizing time-lapse microscopy and single-cell cloning assay, we observed that neosis derived from those PGCCs with proliferative capacity contributed to tumor cell repopulation after irradiation. Using the conditioned media collected from dying tumor cells to perform single-cell cloning assay, we unexpectedly demonstrated that HMGB1 released from dying tumor cells participated the process of neosis-based tumor repopulation. In irradiation treated animal tumor bearing model, the expression level of HMGB1 increased after irradiation compare with non-irradiated group. Moreover, some PGCCs presented high HMGB1 expression. Interestingly, we also observed that the proliferation potential of PGCCs varied. Some PGCCs proliferated at early stage, while some PGCCs proliferated at late stage.
Conclusion X-ray irradiation could induce the formation of PGCCs, which could move towards both cell death and survival; irradiation-generating PGCCs mediated tumor cell repopulation after irradiation via neosis; HMGB1 released from dying cells stimulated the process of neosis and participated in tumor repopulation after irradiation.

Background
Tumor repopulation, a great predicament in cancer therapeutics, generally denotes that surviving tumor cells undergoing therapies (e.g. radiotherapy, chemotherapy) continuously proliferate and unfavorably reestablish the tumor. It is conceivable that unfolding the mechanisms responsible for tumor repopulation may bring novel therapeutic strategies to restrict tumor repopulation and improve clinical outcomes.
Emerging studies have focused on unraveling the mechanisms underlying tumor repopulation after therapy. Huang et al. in 2011 demonstrated that radiotherapy-induced dying tumor cells exerted great potential to stimulate surrounding tumor cells proliferation through a Caspase-3 dependent mechanism [1]. Another successive study recapitulated that Caspase-3 in cytotoxic therapy-induced dying melanoma cells prompted melanoma repopulation after therapy [2]. In addition, our laboratory also revealed that PKCδ, which can be cleaved and activated by cleaved Caspase-3, mediated pancreatic tumor cell repopulation as a downstream factor of Caspase-3 [3]. Besides Caspase-3-centered tumor repopulation mechanisms, one recent study from our laboratory uncovered that eIF4E-phosphorylation-mediated Sox2 upregulation facilitates pancreatic tumor cell repopulation following irradiation [4]. Though these studies have enriched theories and molecular mechanisms of tumor repopulation, the observably morphological mechanisms at cellular level, which may directly display the process of tumor repopulation, are elusively understood.
Importantly, increasing studies have shown that ionizing irradiation can also induce the formation of polyploidy giant cancer cells [11][12][13][14]. The main mechanisms responsible for the formation of PGCCs are associated with cell fusion [15], endoreplication [16,17], cytokinesis failure [16,17], cell cannibalism by entosis [18], eventually contributing to generation of polyploidy cells. Several studies have reported that PGCCs are connected with therapeutic resistance, including chemo-resistance [19,20] and radioresistance [14]. Unfortunately, the precise mechanisms of how these PGCCs contribute to therapy resistance are still vaguely understood. Regarding the fate of PGCCs, some studies reported that PGCCs eventually move towards cell death via mitotic catastrophe [21][22][23].
Neosis, a novel manner of cell division, was rstly reported by Sundaram et al. in 2004 [24]. Before the demise of PGCCs, some of them can undergo neosis characterized by karyokinesis via nuclear budding, asymmetrically intracellular cytokinesis and production of small mononuclear cells (be termed Raju cells) [24]. The newly formed Raju cells via neosis were considered to play a role in self-renewal in cancer [25,26], therapeutic resistance [27,28] due to their stem-like traits [29]. However, the de nite role of neosis in tumor recurrence or repopulation after therapy remains unclear.
In this study, we found that the PGCCs induced by irradiation can move towards both demise and survival. Importantly, using single-cell cloning assay, we observed that those PGCCs with potentially proliferative capacity eventually mediated tumor cell repopulation via neosis. In a further step, we demonstrated that HMGB1 secreted by dying tumor cells promoted neosis through a paracrine manner.
We hope that this study could at cellular and molecular levels provide novel perspective on tumor repopulation after radiotherapy.

Materials And Methods
Cell culture and irradiation The breast cancer cell line 4T1 and MDA-MB-231 cells were purchased from the Chinese Academy of Science (Shanghai, China) and cultured in Dulbecco's Modi ed Eagle's Medium (DMEM) (Thermo Scienti c Inc., Beijing, China) containing 10% fetal bovine serum (FBS) (Tianhang Biological Technology Immunocytochemical and immuno uorescence staining Cells grown on growth cover glasses (Fisher Scienti c, Pittsburgh, PA, USA) positioned on the bottom of 24-well dish or specialized confocal dishes with coverglass bottom (In Vitro Scienti c, CA, USA), were irradiated 48h after seeding. At the indicated time points the cells were xed in paraformaldehyde (4% in PBS, pH 7.4), permeabilized with 0.3% Triton-X100, and blocked with 5% donkey serum in PBS for 1 h at room temperature. The cells were then incubated with primary antibodies at 4°C overnight, followed by incubation with uorochrome-labeled secondary antibodies (Biotium, Hayward, CA, USA) for 20 min at room temperature. For double immuno uorescence, primary antibodies raised in rabbit and mouse species in a mixture, and CF633-labeled anti-rabbit and CF488A-labeled anti-mouse secondary antibodies (Biotium, Hayward, CA, USA) in a mixture were used. Nuclei were counterstained with DAPI (Beyotime, Jiangsu, China), and cells were analyzed using Leica SP8 confocal laser scanning microscope. For immunocytochemistry, GtvisionIII detection system (Gene Tech, Shanghai, China) was used as equivalent as secondary antibody after incubation of primary antibody as previously shown, then substrate diaminobenzidine was added to visualize positive immune reaction. Nuclei were counterstained with hematoxylin.
Terminal deoxyribonucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) Cells grown on cover glasses were xed in paraformaldehyde (4% in PBS, pH 7.4), permeabilized (0.3% Triton X-100 in PBS), and then incubated in the TUNEL reaction mixture (In Situ Cell Death Detection Kit, TMR red, from Roche, America) for 1 hour. The TUNEL positive cells were nally analyzed using Leica SP8 confocal laser scanning microscope.

Transmission electron microscopy
Exponentially growing cells in 10 cm dishes were irradiated with X-ray of 10Gy doses. After 4 days, the adherent cells were xed in 2.5% glutaraldehyde solution overnight, and then scraped off with cell scraper. After centrifugation and washing with PBS, the cells was xed with osmium tetroxide solution for at least 2 hours, and dehydrated with gradient alcohol and propylene oxide. The cell pellet was exposed to mixture of propylene oxide/resin for in ltration and then for embedding. Sectioning is accomplished with the use of an LKB Ultratome V. The thin sections were transferred into a copper grid and then stained with lead citrate for viewing under PHILIP CM-120 transmission electron microscope (Philips, Netherlands).

Brdu immuno uorescence staining and Brdu chasing assay
A nal concentration of 0.03 mg/mL of BrdU (Sigma-Aldrich, Missouri, USA) in pre-warmed growth medium was added to cells grown on coverslips and incubate at 37°C for 30 minutes. The cells were xed in paraformaldehyde (4% in PBS, pH 7.4), then incubated in 1.5M HCl for 30 minutes at room temperature to denature DNA. Brdu primary antibody (Cell Signaling Technology, Beverly, MA, USA) was then used to detect the incorporated Brdu in cells through immuno uorescence staining as shown above.
Brdu pulse-chase assay was performed to track Brdu-labeling PGCCs. Brdu (0.03 mg/mL) was added to cells immediately for one pulse incubation for 6 hours to make most cells Brdu-labeled after the cells were irradiated. At the time of 6h, 2d,5d,10d,15d after labeling, remnant incorporated Brdu were detected by immuno uorescence staining.
Time lapse observation of 4T1 PGCCs following irradiation 4T1 cells seeded in 24 well plate in density of 20000 per well were irradiated with 10Gy doses of X-ray. The morphological change of PGCCs was analyzed by IncuCyte ZOOM Live-Cell Analysis System, two sampling point was selected in every well of the plate, and photo was shot once every 2 hours with total shooting duration of 5 days.
The record of the PGCCs and neosis and single-cell cloning assay The irradiated 4T1 cells were cultured for 4 or 5 days until the largest numbers of PGCCs appear, then the cells were trypsinized and 1000 cells were seeded into 10cm petri dishes for further culture. On the next day the single adherent viable cells were labeled and photographed every 12 hours under phase contrast microscope. To con rm that a signal tumor cell treated like above was able to form clone from PGCC by neosis, a signal tumor cell treated like above was plated into signal well of 96 well plate by FACS sorting. From the next day on the well with adherent cell was marked and observed for 14 days. The plate was by stain crystal violet and colony with more than 50 cells was counted.

Animal models
All of the BALB/c nude mice were brought from Shanghai slack laboratory animal co. LTD (Shanghai, China). For animal tumor bearing model, 1 ×10 6 cells suspended in 100 μL PBS were subcutaneously injected into the hind leg of 6 weeks old BALB/c nude mouse. When tumor volume reached 200m 3 , the nude mice were obtained 10Gy partial irradiation centralized on the tumor. Tumor volume was measured twice per week calculating with formula V= length×width 2 /2. All procedures of animal experiments were performed according to The Animal Care and Use Committee of Shanghai Jiao Tong University School of Medicine.

HE staining
The formalin-xed, para n-embedded spheroid tissue were cut in to 4-micrometer sections, then depara nized and rehydrated. After that, these tissue sections were counterstained with hematoxylin for 1 min and eosin for 2 min respectively, then dehydrated and mounted with coverslips.
Immunohistochemical staining (IHC) The 4-micrometer sections were washed with PBS and then subjected to antigen retrieval in 0.01 M sodium citrate buffer, heated in water bath for 10 min. The sections were incubated with primary antibodies at 4 ℃ overnight in a humidifying chamber. Streptavidinbiotin system was used to detect the signal in the presence of the chromogen 3,3'-diaminobenzidine or alkaline phosphatase.

Statistical Analysis
Student's two-sided test was used to compare the signi cance between control and treated groups. Multiple group comparisons were performed using one-way analysis of variance (one-way ANOVA). P < 0.05 was considered statistically signi cant.

Fate of irradiated tumor cells
We rstly determined the in uence of irradiation on tumor cell fate. Clonogenic cell survival assay showed the survival fraction of 4T1 cells in response to different doses of X-ray irradiation (Fig. 1A). For example, 10 Gy irradiation killed most of 4T1 cells and only approximately 1.6% cells survived and formed colonies. Flow cytometry analysis relying on FITC Annexin V and propidium iodide (PI) double staining displayed that the percentage of 4T1 cells in early apoptosis and necrosis increased signi cantly 4 days after 10 Gy irradiation (Fig. 1B). Western blot analysis manifested that the expression of Cleaved Caspase-3, secreted HMGB1, ratio of LC3II/I rose after 10 Gy irradiation, indicating that multiple cell death manners existed after irradiation, like apoptosis, necrosis and autophagy (Fig. 1C). Transmission electron microscopy con rmed these cell death manner (Fig. 1D).

Irradiation induces the formation of PGCCs
Despite of multiple cell death manners after irradiation, we also observed the morphological change of these cells. Intriguingly, we found that the cell size of some 4T1 cells after irradiation continuously increased and hundreds of multinuclei and micronuclei can be seen even in one single enlarged cell. These cells were named PGCCs (Fig.1E). PGCCs existed after irradiation, and the percentage of these PGCCs grew rapidly in the following days, which reached peak value (42%) at day 5 and decreased in the next days (Fig.1E). To con rm the existence of PGCCs after irradiation, immuno uorescence and electron microscopy analysis were utilized, with DAPI for nucleus and phalloidin for cytoskeleton. The cytoskeleton staining by phalloidin indicated that multiple nuclei located inside one single PGCC (Fig.1F). PGCC also could be clearly seen in transmission electron microscope (Fig.1F). Cell cycle analysis, which identi ed PGCC as polyploid cells, was done to further gure out PGCC changes over time following irradiation under different doses of irradiation. Cell cycle analysis revealed that the percentage of PGCCs (polyploid 4T1 cells) greatly augmented after 10 Gy irradiation, ranging from 6.09% to 34.79% (Fig. 1G). The higher the radiation dose, the more the PGCCs (Fig.1G). These results revealed the multinucleated trait of PGCCs and excluded the possibility that PGCCs were aggregate of many cells or cellular debris after irradiation. The PGCCs induced by irradiation were characterized by signi cantly increased cell size, the multiple nuclei or giant nuclei gathered in the center of cell, and the micronuclei widely scattered around the multiple nuclei.
PGCCs formed after irradiation exhibited proliferation potential To gure out the destiny of PGCCs induced by irradiation, real-time live cell tracing was performed to record the death course of a single PGCC after 10Gy X-ray irradiation using the Incucyte ZOOM Live Cell Imager. The result showed that PGCC began to die gradually at day3, and fully disintegrated at day 5 nally (Fig. 2A). The apoptosis of PGCCs was detected with TUNEL and immuno uorescence for Cleaved Caspase-3 (Fig. 2B). To investigate whether PGCCs had mitotic capacity, Brdu pulse-chase assay was performed. Brdu with full dose was administered to 4T1 cells immediately after X-ray irradiation, so the majority of the cells were able to obtain exogenous Brdu incorporation. The content and distribution of remaining Brdu were then tracked. Although almost all the cells were Brdu positive at 6h, Brdu positivelystaining cells decreased sharply in the next few days, and the green uorescence of Brdu was not observed in the newly formed PGCC (Fig. 3C). The immuno uorescence co-localized the signals of Brdu and γ-H2AX, showing that the proliferative PGCCs indeed experienced DNA double strand injury imposed by irradiation (Fig. 2D). Western blot con rmed that PCNA, a well-known proliferative marker, rose with increased γ-H2AX, a cell damage signal, at 30 min after irradiation, indicating that increased proliferative activity with ionizing damage is a rapid reaction (Fig. 2D). We also performed the staining of proliferative markers Ki-67. Compared to normal non-irradiated cells, Ki-67 index was obviously increased on 4 days after irradiation. Moreover, Ki-67 was demonstrated strong positive in most of PGCCs (Fig. 2E).
To con rm PGCCs existence after irradiation in vivo, we utilized 10Gy irradiation to treat animal tumor bearing model. Irradiation treatment indeed slowed down the rate of tumor growth ( Fig.3A and B). HE staining was used to detect the morphological change of these irradiated tumor cells. Compared to nonirradiation group, PGCCs were observed in irradiation group after 10 days, and increased in the following 4 days. After 21 days, PGCCs were rarely seen in tumor tissue and most tumor cells become normal size as non-irradiation group (Fig.3C). To test whether these PGCCs also presented proliferation potential just like in vitro assay, Brdu and Ki-67 immunohistochemical staining were applied. Some PGCCs were Ki-67 and Brdu positive, indicating that these PGCCs have proliferation potential (Fig.3D).

PGCCs involved in tumor repopulation via neosis
Were some PGCCs have the ability to escape the fate of death because PGCCs were proliferative with mitotic activity? We used single-cell cloning assay to observe the fate of PGCCs. Through time-lapse observation, we found that some PGCCs in the 96-well plate could generate numerous small mononuclear cells (Raju cells, RCs) and eventually form cell colonies, which was termed neosis (Fig.4A). HE staining veri ed the formation of PGCCs and RCs after irradiation (Fig.4B). These RCs might come from PGCCs budding or cell division ( Fig.4C and D). We speculated these newly formed cells through neosis were responsible for tumor cells repopulation. Using western blot analysis, we detected the proliferative ability of RC. Results showed that RCs expressed higher levels of CyclinD1 and PCNA, compared with non-irradiated 4T1 cells and the mixture of PGCCs and RCs and lower level of Cleaved Caspase-3, compared with the mixture of PGCCs and RCs (Fig. 4E), indicating RCs had stronger ability of proliferation and lower level of apoptosis. HMGB1 released from dying cells mediated neosis-initiating tumor repopulation HMGB1 is one of the important damage associated molecular patterns (DAMP) [30], which could be released from apoptotic cells [31] or necrotic cells [32]. We subsequently explored the possible role of HMGB1 in Neosis. Firstly, we utilized FACS to sorting PGCCs formed at 4 days after irradiation and used single-cell cloning assay to observe the destiny of these PGCCs (Fig.5A). This single-cell cloning assay excluded the possibility that the newly formed colonies were aggregate of many cells and it can also measure the ability of neosis from PGCCs. As shown in Fig.5B, some PGCCs indeed formed new cell colonies through cell budding (also named neosis). Since western blot analysis showed that HMGB1 in supernatant increased after irradiation (Fig.1C), we then used the conditioned media from irradiated 4T1 cells to perform the single-cell cloning assay. Fig. 5C showed that the conditioned media from irradiated 4T1 cells could signi cantly increase the colony formation of PGCCs via neosis. More importantly, two HMGB1 inhibitors, ethyl pyruvate and glycyrrhizinate, could remarkably neutralize the single-cell cloning ability of PGCCs via neosis (Fig. 5C), suggesting that the HMGB1 in the conditioned media from irradiated 4T1 cells mediated this process. Additionally, we analyzed the expression of HMGB1 in irradiation treated animal tumor bearing model as well. As shown in Fig.5D, the expression level of HMGB1 increased after irradiation compare with non-irradiated group. Moreover, some PGCCs presented high HMGB1 expression. Interestingly, we also observed some PGCCs with Brdu positive but Ki-67 negative, or Brdu negative but Ki-67 positive, revealing that the proliferation potential of PGCCs varied, in other words, some PGCCs proliferated at early stage, while some PGCCs proliferated at late stage.
In sum, we found that X-ray irradiation could induce the formation of PGCCs, which could move towards both cell death and survival; irradiation-generating PGCCs mediated tumor cell repopulation after irradiation via neosis; HMGB1 released from dying cells stimulated the process of neosis and participated in tumor repopulation after irradiation (Fig. 5E).

Discussion
PGCCs can be observed in both human cancer cells [5][6][7][8][9][10] and induced by cancer therapy [11-14, 19, 20]. Concerning the fate of PGCCs once they formed, some reports found that PGCCs nally come to death through mitotic catastrophe [21][22][23]. Intriguingly, emerging studies have found that PGCCs could form numerous small mononuclear cells, which process is termed neosis [24,25,27]. Also, increasing evidence showed that neosis deriving from PGCCs had a potential role in therapeutic resistance [27,28], which may be undervalued before. Unfortunately, the direct role of neosis in tumor repopulation or recurrence was poorly understood. In this study, we aimed mainly to illustrate the directly-participating role of neosis in tumor cell repopulation after radiotherapy. We rstly found that irradiation-induced PGCCs could eventually come to demise. However, we unexpectedly detected that some of the PGCCs in spite of DNA damage by irradiation still bore proliferative ability. Importantly, those PGCCs with proliferative property were observed to contribute to tumor repopulation through neosis mechanism.
Finally, our results demonstrated that HMGB1 released from dying tumor cells mediated the process of neosis and eventually bring about tumor repopulation.
Our results showed that the ratio of PGCCs following irradiation peaked at day 5 and decreased later (Fig. 1E). The drop in ratio of PGCCs in later days was attributed possibly to death of PGCCs, because there were studies reported the demise fate of PGCCs [21][22][23]. It is also highly possible that repopulation of newly emerging cells through neosis made this ratio drop as well, because we found that PGCCs contributed to tumor repopulation after radiotherapy via neosis ( Fig. 4 and Fig. 5B). Though tumor colony was repopulated 15 days after irradiation, the newly-forming tumor cell repopulation still contained a higher percentage of PGCCs than in the non-irradiated group (Fig. 1E), suggesting the non-negligible role of PGCCs in tumor cell repopulation. Western blot analysis showed that Cleaved Caspase-3 appeared since day 1 after radiation, rose gradually in the following days, reached the peak at day 4 or 5, and declined 6 days after radiation (Fig. 1C). The trend of Cleaved Caspase 3 expression following irradiation was consistent with that of the ratio of PGCCs after radiation (Fig. 1E and G), suggesting that PGCCs may die also through the procedure of apoptosis. As shown in Fig. 1G, larger dose of irradiation induced higher ratio of PGCCs (47.8% versus 16.1%, 14 Gy versus 6 Gy on day 3), whereas those PGCCs induced by larger dose of irradiation were more prone to demise (10.6% versus 23.1%, 14 Gy versus 6 Gy on day 15), suggesting that PGCCs induced by lower dose of irradiation maintained for a longer period of time due probably to less severe DNA damages.
We used Brdu pulse-chase assay to con rm whether the PGCCs possessed proliferation (Fig. 2C). The result of this assay indicated that PGCC formed newly and Brdu incorporated into DNA was attenuated with continuous cell division activity. The results do not support PGCCs as a resting or dormant cell, even though the emerging PGCC in 15 days after irradiation may be newly formed. Besides this assay, we also utilized multiple methods to uncover the proliferative capacity of some PGCCs (Fig. 2D-E and Fig. 3D), unexpectedly nding that in spite of DNA damage, some PGCCs could still bear proliferative potential. To clearly demonstrate the fate of those proliferative PGCCs, we utilized time lapse microscopy to observe the phenomenon of neosis (Fig. 4A-D and Fig. 5B). Fortunately, we precisely captured the process of neosis, in which two small Raju cells are being expelled from the PGCC (Fig. 4C). We observed that the RCs had stronger ability of proliferation and lower level of apoptosis (Fig. 4E), which was consistent with the opinion that neosis mediates tumor progression after therapy [24][25][26] and therapy resistance [27,28].
To quantitatively measure the role of PGCCs-based neosis in tumor repopulation after radiotherapy, we innovatively modi ed the single-cell cloning assay (Fig. 5A and B). Employing this method, we also explored the potential role of HMGB1 in neosis-based tumor repopulation (Fig. 5C).
This study for the rst time revealed the role of HMGB1 in neosis-based tumor repopulation, consolidating and enriching the role of HMGB1 in tumor recurrence after therapy [33] and tumor progression [34,35]. In addition, one previous study in our laboratory also revealed that HMGB1 from dying tumor cells stimulated the proliferation of living tumor cells and presaged poor prognosis in cancer patients, whereas this study did not determine the effect of HMGB1 on neosis and tumor repopulation after radiotherapy [36].

Conclusion
In conclusion, our ndings demonstrated that irradiation-induced PGCCs contributed to tumor repopulation via neosis, in which HMGB1 was involved. We hoped that the theory of neosis-based tumor repopulation we proposed in this study could provide novel insights to understand the process of tumor repopulation after therapy at cellular and molecular levels.