Molecular signature of anastasis for reversal of apoptosis

Apoptosis is a type of programmed cell death that is essential for normal organismal development and homeostasis of multicellular organisms by eliminating unwanted, injured, or dangerous cells. This cell suicide process is generally assumed to be irreversible. However, accumulating studies suggest that dying cells can recover from the brink of cell death. We recently discovered an unexpected reversibility of the execution-stage of apoptosis   and in vitro in , and proposed the term anastasis (Greek for “rising to life”) to describe this vivo cell recovery phenomenon. Promoting anastasis could in principle preserve injured cells that are difficult to replace, such as cardiomyocytes and neurons. Conversely, arresting anastasis in dying cancer cells after cancer therapies could improve treatment efficacy. To develop new therapies that promote or inhibit anastasis, it is essential to identify the key regulators and mediators of anastasis – the therapeutic targets. Therefore, we performed time-course microarray analysis to explore the molecular mechanisms of anastasis during reversal of ethanol-induced apoptosis in mouse primary liver cells. We found striking changes in transcription of genes involved in multiple pathways, including early activation of pro-survival genes, cell cycle arrest, stress-inducible responses, and at delayed times, cell migration and angiogenesis. Here, we present the time-course whole-genome gene expression dataset revealing gene expression profiles during the reversal of apoptosis. This dataset provides important insights into the physiological, pathological, and therapeutic implications of anastasis. Ho Man Tang ( ), Ming Chiu Fung ( ), Ho Lam Tang ( Corresponding authors: homantang@jhmi.edu mingchiufung@cuhk.edu.hk ) holamtang@jhmi.edu Tang HM, Talbot Jr CC, Fung MC and Tang HL. How to cite this article: Molecular signature of anastasis for reversal of apoptosis [version 2017,  :43 (doi: ) 1; referees: 1 approved, 1 approved with reservations] F1000Research 6 10.12688/f1000research.10568.1 © 2017 Tang HM  . This is an open access article distributed under the terms of the , which Copyright: et al Creative Commons Attribution Licence permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. This work was supported by the Life Science Research Foundation fellowship (H.L.T.). Grant information: The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: No competing interests were disclosed. 13 Jan 2017,  :43 (doi: ) First published: 6 10.12688/f1000research.10568.1 1 1 2


Introduction
Apoptosis (Greek for "falling to death") was generally assumed to be an irreversible cell suicide process because it involves rapid and massive cell destruction [1][2][3][4][5][6][7] . During apoptosis, intrinsic and extrinsic pro-apoptotic signals can converge at mitochondria, leading to mitochondrial outer membrane permeabilization (MOMP), which releases cell execution factors, such as cytochrome c to trigger activation of apoptotic proteases including caspase-3 and -7 8,9 , small mitochondria-derived activator of caspases (Smac)/direct IAP binding protein with low pI (DIABLO) to eliminate inhibitor of apoptosis protein (IAP) inhibition of caspase activation 10,11 , and apoptosis-inducing factor (AIF) and endonuclease G to destroy DNA 12-15 . Activated caspases commit cells to destruction by cleaving hundreds of functional and structural cellular substrates 2,16 . Crosstalk between signalling pathways amplify the caspase cascade to mediate cell demolition via nucleases (DNA fragmentation factor [DFF]/caspase-activated DNase [CAD]) to further destroy the genome 17-19 , and alter lipid modifying enzymes to cause membrane blebbing and apoptotic body formation 20,21 . Therefore, cell death is considered to occur after caspase activation within a few minutes 22,23 .
However, we and other groups have demonstrated reversal of early stage apoptosis, such as externalization of phosphatidylserine (PS) in cultured primary cells and cancer cell lines [24][25][26][27] . We have further demonstrated that dying cells can reverse apoptosis even after reaching the generally assumed "point of no return", such as MOMP-mediated cytochrome c release, caspase activation, DNA damage, nuclear fragmentation, and apoptotic body formation 26-28 . Our observation of apoptosis reversal at late stages is further supported by an independent study, which shows recovery of cells after MOMP 29 . To detect reversal of apoptosis in live animals, we have further developed a new in vivo caspase biosensor, designated "CaspaseTracker" 30 , and successfully identified and tracked somatic, germ and stem cells to survive transiently-induced cell death, and potentially during normal development and homeostasis in Drosophila melanogaster after caspase activation 30,31 , the hallmark of apoptosis 2,32 . We refer to this recovery phenomenon as "anastasis" 27 , which means "rising to life" in Greek, for the reversal of apoptosis. Anastasis appears to be an intrinsic cell survival phenomenon, as removal of cell death stimuli is sufficient to allow dying cells to recover [26][27][28]30 .
The physiological, pathological and therapeutic importance of anastasis is not yet known. We proposed that anastasis could be an unexpected tactic that cancer cells use to escape cancer therapy 26-28 . Many tumours undergo dramatic initial responses to cell death-inducing radiation or chemotherapy 33-36 ; however, these cells relapse, and metastasis often occurs in most types of cancer [33][34][35] . Therefore, the ability of cells to recover from transient induction of cell death may allow tumour cells to escape treatment, and survive and proliferate, resulting in relapse [26][27][28] . Furthermore, cells may acquire new oncogenic mutations and transformation phenotypes during anastasis 27,28 , such as DNA damage caused by apoptotic nucleases. Therefore, anastasis could be one mechanism underlying the observation that repeated tissue injury increases the risk of cancer in a variety of tissues 37 , such as liver damage due to alcoholism 38 , chronic thermal injury in the oesophagus induced by the consumption of very hot beverages 39-41 , evolution of drug resistance in recurrent cancers [26][27][28] , and development of a second cancer during subsequent therapy [42][43][44][45] . Anastasis can also occur in primary cardiac cells and neuronal cell lines 27,28 , and potentially in cardiomyocytes in vivo following transient ischemia 46 . These findings suggest anastasis as an unexpected cellular protective mechanism. Therefore, uncovering the mechanisms of anastasis may provide new insights into the regulation of cell death and survival, and harnessing this mechanism via suppression or promotion of anastasis would aid treatment of intractable diseases including cancer, heart failure and neurodegeneration.
Our previous study demonstrated reversibility of ethanol-induced apoptosis at late stages in mouse primary liver cells, and revealed that new transcription is important to reverse apoptosis 27,28 . During recovery, there was up-regulation of genes involved in pro-survival pathways and DNA damage responses during anastasis (Bag3, Mcl1, Dnajb1, Dnajb9, Hsp90aa1, Hspa1b, and Hspb1, Mdm2) 27 . Interestingly, inhibiting some of those genes by corresponding specific chemical inhibitors significantly suppresses anastasis 27 . However, the molecular mechanism of anastasis remains to be elucidated. To study the cellular processes of anastasis, we performed time-course RNA microarray analysis to determine the gene expression profiles of cultured mouse primary liver cells undergoing anastasis following exposure to ethanol, and identified unique gene expression patterns during reversal of apoptosis. Here, we present our time-course microarray data, which reveals the molecular signature of anastasis.

RNA isolation, reverse transcription, and microarray
Mouse primary liver cells were isolated from BALB/c mice using collagenase B and cultured as described 27,47 . The cells were treated with 4.5% ethanol in DMEM/F-12 (DMEM:nutrient mixture F-12) supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Carlsbad, CA, USA) at 37°C under an atmosphere of 5% CO 2/95% airfor 5 hours (R0), and then washed and further incubated in fresh culture medium for 3 hours (R3), 6 hours (R6), 24 hours (R24), and 48 hours (R48). Three biological replicates were performed at each time point. The untreated cells served as control (Ctrl). Total RNA in the corresponding cell conditions was harvested using TRIzol Reagent, and RNA was purified using the RNeasy Mini Kit (Qiagen, Cologne, Germany). Reverse transcription was performed using SABiosciences C-03 RT 2 First Strand Kit to construct cDNA (SABiosciences-Qiagen, Frederick, MD, USA). The cDNA samples were analysed using the Illumina MouseWG-6 v2.0 Expression BeadChip (Illumina, San Diego, CA, USA).

Gene expression data analysis
The Partek Genomics Suite 6.6 (Partek, St. Louis, MO, USA) was used for principal component analysis 48 . Spotfire DecisionSite 9.1.2 (TIBCO, Palo Alto, CA, USA) platform was used to evaluate the fold change of gene expression levels between time points when compared with a common starting point 49 . Signal values were converted into log 2 space and quality control tests were performed to ensure data integrity by comparing the signals of the three biological replicates at each time point. The fold change was based on averaged values of the three replicates at each time point; two-sample Student's t-test was used to determine statistical significance as p-values of less than 0.05, using Partek Genomics Suite v6.5 (Partek Inc., St. Louis, MO, USA).
For the time-course gene expression analysis using Spotfire, all time points were compared with time point Ctrl, which represents untreated cells. Spotfire was used to show the genes that displayed specific changes in gene expression after removal of apoptotic inducers for 3 hours and 6 hours, as well as the genes that were up-regulated from apoptosis (R0) to 6 hours (R6) after removal of the inducer. Genes with specific and significant change (Log 2 > 1 or <-1) in expression at the corresponding timepoint are highlighted. Interaction network analysis of the up-regulated genes during anastasis was performed using GeneMANIA database (http://genemania.org/) 50,51 .

Results and discussion
We have demonstrated that mouse primary liver cells can reverse the apoptotic process at the execution stage, despite experiencing important checkpoints commonly believed to be the "point of no return", including caspase-3 activation, DNA damage, and cell shrinage 27,28 . To pursue the mechanisms of anastasis, we performed time-course high-throughput microarray to evaluate gene expression profiles during reversal of ethanol-induced apoptosis in mouse primary liver cells. RNA samples were collected from the untreated primary liver cells (Ctrl), the cells treated with 4.5% ethanol for 5 hours when cells exhibited hallmarks of apoptosis (R0), and the treated cells that were then washed and cultured in fresh medium for 3 (R3), 6 (R6), 24 (R24) and 48 (R48) hours. Apoptosis was confirmed previously in the ethanol-treated cells (R0), which displaced hallmarks of apoptosis, including plasma membrane blebbing, cell shrinkage, cleavage of caspase-3 and its substrates, such as PARP and ICAP ( Figure 1A  Genes that display significant changes in expression during anastasis at the earliest time point of 3 hours, following the removal of the apoptotic inducer, may represent critical first responders of anastasis ( Figure 3A, Table 1), including transcription factors of the activator protein-1 (AP-1) family (Atf3, Fos, Fosb, Jun, Junb), transforming growth factor-β (TGF-β) signal pathway and its related regulators (Inhba, Snai1, Tgif1, Sox4, Klf4, Klf6, Klf9), prosurvival Bcl-2 family member (Bag3), inhibitor of p53 (Mdm2), anti-proliferation (Btg1), DNA damage (Ddit3, Ddit4) and stressinducible (Dnajb1, Dnajb9, Herpud1, Hspb1, Hspa1b) responses. Starting at 6 hours of anastasis, other groups of gene pathways displace the peak of transcription, such as cell cycle arrest (Cdkn1a, Trp53inp1), autophagy (Atg12, Vps37b), and cell migration (Mmp10 and Mmp13) ( Figure 3B, Table 2 and Table 3). Expression of potent angiogenic factors, such as Vegfa and Angptl4, peaks at 3 and 6 hours of anastasis, respectively. Changes in expression of most of these genes peak at the 3-6-hour time points after removal of the apoptotic stimulus and then return to baseline ( Figure 3A      The change in transcriptional profiles during anastasis provides us mechanistic insights into how dying cells could reverse apoptosis (Figure 4). In early anastasis (R3), our microarray data reveals that the regulators of the TGF-β signalling pathway, which control various fundamental cellular process, including proliferation, cell survival, apoptosis and transformation 53-55 , are upregulated. The activation of the TGF-β pathway is further supported by the upregulation of AP-1 (Jun-Fos) during early anastasis. The up-regulation of the TGF-β pathway also promote the expression of murine double minute 2 (Mdm2) 56,57 , an inhibitor of p53 that is also up-regulated during early anastasis 27 . As p53 plays a critical role in regulating apoptosis and DNA repair 58,59 , the expression of Mdm2 could not only promote cell survival by inhibiting p53mediated cell death, but also cause mutations as we have observed in the cells after anastasis 27 . Expression of Mdm2 can also activate XIAP 60 , which inhibits caspases 3, 7 and 9 61-66 , and therefore, could promote anastasis by suppressing the caspase-mediated cell destruction process. Up-regulation of anti-apoptotic BCL2 protein (Bag3) and heat shock proteins (Hsps) during anastasis can also neutralize pro-apoptotic proteins to promote cell recovery 67-69 . Notably, Bbc3 is a pro-apoptotic BH3-only gene to encode PUMA (p53 upregulated modulator of apoptosis) 70,71 . Its expression peaks at anastasis (R3-R6), suggesting the sign of anastasis vs apoptosis in the recovering cells during the early stage of the cell recovery process.
To reverse apoptosis, the recovering cells need to remove or recycle the destroyed cellular components, such as the toxic or damaged proteins that are cleaved by caspases, and dysfunctional organelles like the permeabilized mitochondria. Autophagy could contribute to anastasis, as the recovering cells display up-regulation of Atg12 ( Figure 3B, Table 2), which is important to the formation of autophagosome to engulf the materials that are then transported to lysosomes or vacuoles for degradation [72][73][74][75] . Recently studies reveal that autophagy can be activated by the DNA damage response, and play a role in maintaining the nuclear and mitochondrial genomic integrity through DNA repair and removal of micronuclei and damaged nuclear parts 76,77 . This could suppress mutagenesis and oncogenic transformation to occur in the cells that reverse apoptosis as observed after DNA damage 27,28 . Autophagy is also implicated in the exosome secretory pathway 78-80 , which could allow rapid clearance of damaged or toxic materials during anastasis through exosomes. Interestingly, our microarray data shows that the recovering cells display up-regulation of potent angiogenic factors such as Vegfa and Angptl4 ( Figure 3A and B, Table 1 and Table 2), which promote vascular permeability and angiogenesis 81-84 . This could facilitate anastasis by supplying nutrient and clearing waste products. However, this could also enhance tumour progression and metastasis when anastasis occurs cancer cells.
In fact, our data also reveals the up-regulation of genes involved in cell migration during anastasis 27 , such as Mmp 10 and 13 that encode matrix metalloproteinases 85-88 . This could be a stressinducible response that promotes cell migration, like what was observed in HeLa cells after anastasis 28 , which might contribute to wound healing, or metastasis during cancer recurrence 89,90 .
Arresting cell cycle during anastasis is important as it can allow damaged cells to be repaired before they restore proliferation. This hypothesis is supported by the microarray data that reveals up-regulation of genes that suppress cell cycle ( Figure 3A-C). For example, B-cell translocation gene 1 (Btg1) is an anti-proliferative gene 91,92 , which is up-regulated during the early anastasis (R3). At later stage of anastasis (R6), other cell cycle inhibitors express, including Cdkn1a which encodes p21 that induces cell cycle arrest and senescence 93-95 , and also Trp53inp1 which encodes tumor protein p53-inducible nuclear protein 1 that can arrest cycle independent to p53 expression 96 . These suggest that cell cycle is suppressed by multiple pathways during anastasis.  We also identified genes that are up-regulated both during apoptosis and anastasis, such as Gadd45g, and Rnu6 ( Figure 3C, Table 4). Gadd45g functions in growth arrest and DNA repair 97,98 , and therefore, could be the cytoprotective mechanism that preserves the dying cells during cell death induction (R0), and promotes the injured cells to repair when environment is improved (R3 and R6). Rnu6 encodes U6 small nuclear RNA, which is important for splicing of a mammalian pre-mRNA 99-102 . Upregulation of Rnu6 from R0 to R6 suggests that post-transcription regulation could involve during apoptosis and anastasis. In fact, translational regulation also contributes to anastasis. For example, caspase-3, PARP and ICAP are cleaved in dying cells during apoptosis, and the non-cleaved form of corresponding proteins restores after anastasis ( Figure 1B). Interestingly, the mRNA level of caspase-3 and PARP did not show significant increase (see Data availability 52 ). This suggests translational regulation to occur during anastasis.  Our study provides new insights into the mechanisms and consequences of anastasis ( Figure 5) Researchers can analyse our microarray data to further identify the hallmarks of anastasis, understand its role, elucidate molecular mechanisms that reverse apoptosis, and develop therapeutic strategies to control anastasis.
To identify the genes that displace specific change on a transcriptional level, software such as Spotfire can be used to view the gene expression pattern at different time points during the reversal of apoptosis 49 . To study the molecular mechanism of anastasis, Ingenuity Pathway Analysis can be used to create mechanistic hypotheses according to the transcriptional profile 103 . To identify drugs that modulate anastasis, Connectivity Map can be used to identify small molecules that promote or suppress anastasis based on its gene expression signature 104,105 . Anastasis could be a cell survival phenomenon mediated by multiple pathways [26][27][28]30 , so by comparing the gene expression profiles, researchers can study its potential connection to other cellular processes, such as antiapoptotic pathways, autophagy, and stress-inducible responses 75,106-110 . By searching the molecular signature of anastasis, researchers can study the potential contribution to physiopathological conditions, such as metastasis during cancer recurrence, recovery from heart failure and wound healing 89,90,111 . Further data analysis will stimulate generation of hypotheses for future studies involving anastasis. As our understanding of anastasis mechanism expands, it will uncover its potential impacts on physiology and pathology, and offer exciting new therapeutic opportunities to intractable diseases by mediating cell death and survival ( Figure 6). The manuscript by Tang et al. was focused on the elucidation of the molecular mechanisms of an important phenomenon, anastasis, through time-course expression profiling. Anastasis was recently discovered and has not been fully studied yet. It's molecular basis remains to be uncovered. The study provided useful information to better understand this underexplored process. Overall, the experiment was well designed. The time course experiment included six time points, untreated samples as the control, toxin-induced apoptosis, and four time points after removal of toxin. Three biological replicates were performed at each time point. Figure 1 illustrated the experimental design very well. The biological interpretation of microarray results is reasonable. The reviewer has no major concerns. However, several minor changes are needed, especially for the presentation of figures, which could be improved.
First, no multiple test correction was mentioned in the microarray analysis section. It was described that the p-value less than 0.05 was used to declare statistical significance. The reviewer would suggest the authors confirm that. A false discovery rate (FDR) method is needed for multiple test correction.
Second, the PCA result from Figure 2A showed that three biological replicates were closely clustered, which showed a good repeatability. However, the goal of PCA is not just check the repeatability of three replicates of each group (time point). PCA can be also used to examine the relationship among groups. My recommendation is that the authors provide more description for the PCA result. In addition, in Figure  2A, the percentage of PC2 explaining total variation was masked. But based on the value of PC3, it should be greater than 5.89%. Given the high value of PC1, I would suggest plotting a two dimension PCA plot to display the result or re-plotting this three-dimension plot.
Third, it would be useful to list the number of significant differential expression for each comparison. And I guess the clustering result in Figure 3 presented all significant genes. Editorial comments: In the Abstract, "whole genome" can be replaced by "genome-wide".
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed. This study unravels the gene regulatory network that seems to be involved in the process of anastasis. It is interesting that the authors found various genes that appear to participate in ethanol-induced anastasis, suggesting that the dynamic reconstitution of gene regulatory networks might be a prerequisite for rescuing cells from the brink of cell death. Overall, this work is worth being indexed. However, I would like to see the following points in the research addressed, before approval: Anastasis is a developing concept rather than an established one. It would be better to show the expression dynamics of caspase-3, PARP, and ICAD at all analyzed time points (Cont, R0, R3, R6, R24, and R48). In addition, why don't the authors show apoptotic DNA fragmentation to make sure that all the analyzed cells in the anastasis stage definitely underwent apoptosis?
I may have missed noting this, but there is no statistical analysis of the gene expression changes observed in the microarray data. In Fig. 2B, the expression levels of several genes seem different in the same time point replicates. It would be better to show the genes that were induced or suppressed during anastasis, along with the statistical significance of the differences.
Given the importance of understanding the mechanism of anastasis, it would be better to verify the data obtained from microarray analysis, by using quantitative PCR or Western blotting.
I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
No competing interests were disclosed.

Competing Interests:
Author Response 03 Feb 2017 , Johns Hopkins University, USA

Ho Lam Tang
We thank for the enthusiasm and valuable input from the reviewer, and have made the following changes: We have included the Western blot data ( Figure 1B), which shows that caspase-3, PARP and ICAD were cleaved during apoptosis, but then recovered to their original level at 24 hours after removal of the cell death stimulus. Interestingly, our microarray data shows that their level of mRNA remained no significant change at all time points (3, 6, 24 and 48 hours) after removal of the cell death stimulus, compared with the untreated (control) cells (data available at figshare, please see in the manuscript), suggesting the Data availability recovery of corresponding proteins is contributed by the regulation of translation during and after anastasis. The related data and discussion are included in our revised manuscript.
Our earlier studies using time-lapse live cell microscopy and comic assay demonstrated that the current apoptotic induction (4.5% ethanol, 5 hours) can trigger DNA damage. After 2.
3. the current apoptotic induction (4.5% ethanol, 5 hours) can trigger DNA damage. After removal of the stimulus, major of the dying cells can recover. Interestingly, some cells that reversed apoptosis display chromosomal abnormality and oncogenic transformation, indicating reversibility of apoptosis after DNA damage. In our current study, we further found significant reduction of mRNA level of multiple histone genes during anastasis. Notably, cellular levels of histones reduce in response to DNA damage, as to enhance DNA repairing. Therefore, reduction of expression of histones during anastasis could be a sign of cells that recover from DNA damage after apoptosis.
We have included supplementary data with corresponding p-value for statistical significance of fold change for all of the 3 biological replicants of each gene (see Data availability). The software for the microarray data analysis is mentioned at the "Materials and methods" section.
We have verified our data by RT-PCR in human liver cancer HepG2 cell line, and included the data in the new Figure 4.
No competing interests were disclosed. Competing Interests: