Replicative stress and alterations in cell cycle checkpoint controls following acetaminophen hepatotoxicity restrict liver regeneration

Abstract

Objectives

Acetaminophen hepatotoxicity is a leading cause of hepatic failure with impairments in liver regeneration producing significant mortality. Multiple intracellular events, including oxidative stress, mitochondrial damage, inflammation, etc., signify acetaminophen toxicity, although how these may alter cell cycle controls has been unknown and was studied for its significance in liver regeneration.

Materials and methods

Assays were performed in HuH‐7 human hepatocellular carcinoma cells, primary human hepatocytes and tissue samples from people with acetaminophen‐induced acute liver failure. Cellular oxidative stress, DNA damage and cell proliferation events were investigated by mitochondrial membrane potential assays, flow cytometry, fluorescence staining, comet assays and spotted arrays for protein expression after acetaminophen exposures.

Results

In experimental groups with acetaminophen toxicity, impaired mitochondrial viability and substantial DNA damage were observed with rapid loss of cells in S and G2/M and cell cycle restrictions or even exit in the remainder. This resulted from altered expression of the DNA damage regulator, ATM and downstream transducers, which imposed G1/S checkpoint arrest, delayed entry into S and restricted G2 transit. Tissues from people with acute liver failure confirmed hepatic DNA damage and cell cycle‐related lesions, including restrictions of hepatocytes in aneuploid states. Remarkably, treatment of cells with a cytoprotective cytokine reversed acetaminophen‐induced restrictions to restore cycling.

Conclusions

Cell cycle lesions following mitochondrial and DNA damage led to failure of hepatic regeneration in acetaminophen toxicity but their reversibility offers molecular targets for treating acute liver failure.

1. INTRODUCTION

Acetaminophen (APAP) is widely available over‐the‐counter as an analgesic and antipyretic agent. However, its margin of safety is relatively small, and APAP over ingestion is the leading cause of acute liver failure (ALF) in USA and Western Europe.1 Despite the capacity of the liver to regenerate, it fails to do so in ALF, which causes substantial morbidity and mortality. Although some hepatocytes exhibit DNA synthesis in ALF,2, 3 it has been unclear as to why this is infrequent.

Generally accepted APAP toxicity mechanisms concern production of highly reactive metabolites, eg, NAPQI, with oxidative events and processes, including mitochondrial dysfunction, protein modifications, enzymatic DNA cleavage, engagement of cell death receptors with apoptosis, necrosis or necroptosis, etc.4, 5, 6, 7 The contribution of mitochondria in APAP‐induced oxidative stress has been emphasized.8 Some people with APAP‐induced ALF may recover after intensive care,1 suggesting hepatic damage could be reversible, although residual hepatocytes must re‐enter or traverse the cell cycle. But in an older report, APAP was found to inhibit mitosis in blood cells,9 and in another, to inhibit DNA synthesis in renal cells.10 Also, APAP altered growth factor response, such that EGF receptors, which normally activate proliferation events, were translocated instead to mitochondria and exerted cytotoxic effects.11 Superimposition of genotoxicity may perturb cell cycling events, eg, the mutagen 2‐acetaminofluorene (2‐AAF) inhibited partial hepatectomy (PH)‐induced liver regeneration (LR) by activating the cell cycle suppressor, TP53, and its downstream regulator, CDKN1A (p21).12 Importantly, TP53 and CDKN1A were found to subserve effects of ataxia telangiectasia‐mutated (ATM) gene.13 This ATM signalling is critical for DNA damage response (DDR) and was involved in chemotherapeutic, as well as APAP toxicity.14, 15, 16 The effect of ATM extended to LR since outcomes after PH worsened in ATM knockout mice.17 Moreover, ATM signalling is vital for both advancing and restricting cell cycling in a context‐specific manner.18

Another major aspect of cell cycle regulation concerns mitochondria. For instance, in mitochondrial DNA depletion states due to genetic insufficiencies of deoxyguanosine kinase (DGUOK), DNA polymerase subunit gamma (POLG), mitochondrial inner membrane protein (MPV17), or other genes, ALF develops early in life.19, 20, 21 Mitochondrial DNA may be depleted in liver disease or injury, and may contribute in drug‐induced ALF, eg, following valproate or isoniazid toxicity.22 The critical role of mitochondrial biogenesis and fission in cell cycle progression and cell division has also gathered attention.23 Activation by G1 and G2 cyclins and their partners of dynamin‐related protein (Drp)‐1 in drosophila with its human counterpart, dynamin‐1 like (DNM1L) and mitochondrial fission 1 protein (FIS1), is essential for mitochondrial fission and fusion, without which mitosis is arrested. Similarly, the importance of ATM and its partners, eg, STAT3, within mitochondria for their functional integrity and clearance via mitophagy has been emphasized.24, 25 Therefore, cell cycling mechanisms involving mitochondrial deficits and ATM‐related DDR should be highly significant for LR in APAP toxicity.

Primary hepatocytes are generally insufficient for cell cycling assays due to their restriction in G0/G1 and inability to proliferate in culture conditions. To overcome this difficulty, we used HuH‐7 cell line that was derived from human hepatocellular carcinoma, and has been useful for drug studies.15, 16, 26 We established aspects of APAP toxicity in HuH‐7 cells, including mitochondrial dysfunction, DNA damage and cell death, followed by studies of cell cycle regulation. To address disease‐relevance of findings in HuH‐7 cells, we verified results in primary human hepatocytes and examined hepatic explants during liver transplantation for APAP‐induced ALF. We found APAP toxicity led to DNA damage‐related perturbations in cell cycle checkpoints, S phase entry and cell cycle restriction or even exit from the cycle in surviving cells. However, these events and processes were reversible as shown by studies with granulocyte‐colony stimulating factor (GCSF), which protected cells from drug‐induced DNA damage.15 These cell proliferation mechanisms will be helpful for restoring LR in APAP hepatotoxicity.

2. MATERIALS AND METHODS

2.1. Human liver samples

Anonymized frozen specimens and paraffin‐embedded sections from explanted healthy donor livers (n = 3) and APAP‐induced ALF (n = 6) were from Liver Tissue Procurement and Distribution Service of NIDDK (University of Minnesota School of Medicine, Minneapolis, MN). Tissue procurement was approved by IRB at Albert Einstein College of Medicine.

2.2. Chemicals and drugs

Reagents and APAP were from Sigma‐Aldrich (St Louis, MO). Thiazolyl blue dye (MTT) was from Gold Bio (St. Louis, MO). Stocks for APAP were prepared in dimethylsulphoxide. Recombinant GCSF (filgrastim) was purchased from local pharmacy (Neupogen; Amgen Inc., Thousand Islands, CA).

2.3. Cell culture and cytotoxicity assays

Primary human hepatocytes from healthy donor livers were from ThermoFisher Scientific (Fredericks, MD). HuH‐7 cells were from the Core facility of Liver Center at Einstein and had been authenticated by genotype analysis. Mycoplasma contamination was excluded by PCR and cytostaining methods. Cells were maintained at 37°C in 5% CO₂ in RPMI‐1640 medium (GIBCO, Grand Island, NY) with 10% foetal bovine serum and antibiotics. For cytotoxicity assays, 2‐5 × 10⁴ cells were plated per well in 24‐ or 48‐well dishes with or without APAP or additives. Cell viability was determined by MTT dye reduction as described previously.27 All conditions were in triplicate and experiments were repeated.

2.4. DNA damage assays

For double‐strand DNA breaks, HuH‐7 cells were cultured in glass chamber slides and fixed with acetone. Cells were permeabilized in 0.03% triton‐X for 10 minutes, blocked for 1 hour in 5% goat serum, and incubated with γH2AX antibody (1:100; 2595, Cell Signaling Technologies, Beverly, MA) overnight at 4°C. This was followed by anti‐rabbit Alexa Flour‐647 (1:500; 4414, Cell Signaling Technologies) for 1 hour at room temperature with DAPI (2‐(4‐amidinophenyl)‐1H‐indole‐6‐carboxamidine) counterstaining. γH2AX‐positive fractions were counted in 20 fields each under x200 magnification. Comet assays were as described previously.15 Comet tail length was analysed in 30‐50 cells per condition by CometAssay IV software (Preceptive Instruments, Bury St Edmunds, UK).

2.5. JC‐1 assay for mitochondrial membrane potential (MMP)

HuH‐7 cells were incubated with JC‐1 dye (Cayman Chemical, Ann Arbor, MI), according to manufacturer's instructions, for 1 hour at 37°C. JC‐1 aggregation was recorded by epifluorescence shift from green (monomers) to red (aggregates). Ratio of green and red per cell was determined in replicate images with 300‐500 cells each by Cytation5 instrument (BioTek, Winooski, VT).

2.6. Flow cytometry

Cells were released from dishes by trypsin‐EDTA and kept in 75% ethanol overnight at 4°C. After RNAseA treatment, DNA was stained by 0.1% propidium iodide (PI) (Thermo Fisher Scientific). Typically, 10 000 events were collected by FACS Aria III flow cytometer and cell cycle profiles were obtained by FACS Diva software, v6.1.3 (BD Biosciences Pharmingen, San Diego, CA). For synchronization in G1/S, HuH‐7 cells were cultured overnight with 0.5 mM hydroxyurea (HU), after dose titrations.

2.7. Cell cycle protein expression

HuH‐7 cells and liver tissues were lysed in buffer containing protease and phosphatase inhibitors (Roche Diagnostics, Basel, Switzerland). Total protein content was quantify by Bradford assay and equal amounts of proteins were analysed in duplicate with arrays containing 4 to 6 spots for each of 60 probes (ACC058, Cell Cycle Antibody Array; Full Moon Biosystems, Sunnyvale, CA), according to manufacturer's instructions. After background correction, mean signal intensities were obtained (Genepix; Molecular Devices LLC, Sunnyvale, CA). Protein expression was normalized to β‐actin signals in each sample. We considered ≥ 2‐fold up or downregulation of proteins was significant. Proteins of relevance based on canonical cell cycle events were chosen according to Ingenuity Pathway Analysis (IPA) (Qiagen, Redwood City, CA). Pathways related to ATM‐induced TP53 and CDKN1A were of particular interest. Protein expression differences in networks of interest were analysed by paired Mann‐Whitney rank sum tests.

2.8. Tissue immunostainings

Paraffin‐embedded sections of human livers were deparaffinized with antigen retrieval in citrate buffer. Sections were permeabilized in 0.03% triton‐X‐100 for 10 minutes in PBS, blocked for 1 hour in 5% goat serum, and incubated with antibody to γH2AX (1:100;. 2595, Cell Signaling), p21 (1:100, SC‐379, Santa Cruz Biotechnology, Santa Cruz, CA) or Ki‐67 (1:100; 1:100, 550609, BD Biosciences, San Diego, CA) overnight at 4°C. This was followed by anti‐rabbit or anti‐mouse Alexa Flour‐647 (1:500, 4410 or 4414, Cell Signaling for 1 hour at room temperature with DAPI counterstaining.

2.9. Ploidy class distributions

Nuclei in ethanol‐fixed primary human hepatocytes after overnight culture or in formalin‐fixed tissues from people with ALF were stained by 10 μg/mL Hoechst 33258 dye. Multiple microphotographic images were obtained under x200 magnification and analysed by Cytation5 instrument for nuclear sizes and DNA content per nucleus in hepatocytes. These populations were distributed in comparison with diploid DNA (2N) standards into 2N, 4N, 8N+ and intermediate classes. Ploidy class distributions were separated by Excel software (Microsoft Corp., Seattle, CA), with charting and analysis as follows.

2.10. Statistical analysis

Data are reported as means ± SEM. Statistical significances were analysed by t test for pairwise comparisons, χ² with Fisher's exact test for 4‐fold tables, Spearman correlations for paired data, two‐way ANOVA with Tukey's or Dunnet's post hoc tests for multiple comparisons, or Wilcoxon's rank sum test for ploidy distributions with GraphPad Prism7 software (GraphPad Software Inc., La Jolla, CA). P < .05 was considered significant.

3. RESULTS

In 50% inhibitory concentration (IC50) of 10 mM, APAP cytotoxicity was apparent in HuH‐7 cells by MTT assays, which confirmed losses of cells within 4 hours as well as after overnight (ON) culture (Figure 1A). Drug toxicity produced oxidative stress and MMP declined significantly (Figure 1B). These MMP assays showed mitochondrial injury within 4 hours and also after ON culture, such that JC‐1 dye multimer/monomer ratios decreased to 1.1 ± 0.3 and 0.3 ± 0.1 with APAP vs 2.5 ± 0.2 in controls, respectively, P < .05. Moreover, in APAP‐treated cells, DNA double‐strand breaks appeared, and γH2AX was expressed 4 hour and 24 hours after APAP in 27 ± 1 cells per high power field (hpf) and 28 ± 2 cells per hpf vs 0.7 ± 0.3 cells per hpf in controls, respectively, P < .001 (Figure 1C). Comet assays verified this DNA damage (Figure 1D). Comets were noted after 4 hours in 2.1 ± 0.6% controls vs 28.8 ± 1.9% APAP‐treated cells, P < .001. During ON culture, comets formed in 1.1 ± 0.4% controls vs 70.0 ± 3.0% APAP‐treated cells, P < .001.

Figure 1.

Cytotoxicity of APAP in HuH‐7 cells. (A) Cell viability decreased after HuH‐7 cells were cultured with 10 mM APAP for either 4 h or 16‐18 h (ON). (B) Oxidative stress after APAP culminated in lower MMP with decreased JC‐1 dye aggregates (red) vs monomers (green). This was obvious within 4 h. (C) γH2AX staining (red) for DNA double‐strand breaks. Nuclei were counterstained by DAPI (blue). (D) Comet assay for DNA double‐strand breaks. Comet tails were longer after APAP. Original magnification, B‐D, x200. P values vs controls, ANOVA. Studies were repeated thrice

3.1. APAP rapidly depleted HuH‐7 cells in S and G2/M

Flow cytometry indicated that within 4 hours after APAP, cells in S were depleted by 50% and G2/M by 30% (Figure 2). Cells in S and G2/M were depleted further in continued culture. Moreover, during that period, unlike control HuH‐7 cells, which continued to cycle, APAP‐treated cells exited the cell cycle, as indicated by their enrichment in G0/G1.

Figure 2.

APAP toxicity disrupted cycling in HuH‐7 cells. (A) Cell cycle stages after 4 h or ON culture with APAP. Y‐axis, cell number; X‐axis, DNA content. (B) Quantification for cell cycle subpopulations. Cells in S and G2/M were lost rapidly by 4 h. Cells remained in G0/G1 during ON culture with APAP. P values were calculated by ANOVA. Results were reproduced more than thrice

The susceptibility of cells in S and G2/M to APAP was verified in HuH‐7 cells synchronized in G1/S with HU. Within 4h after release from HU‐induced G1/S arrest, HuH‐7 entered S. By contrast, in cells released from HU arrest, culture with APAP led to depletion of up to 61% in S (Figure 3). This indicated cell survival decreased due to APAP‐induced replicative stress. On the other hand, when APAP was removed after 4 hours and cells were cultured further with fresh medium, cycling normalized. Therefore, cell cycle restrictions imposed by APAP were reversible. However, when HuH‐7 cells were kept ON with APAP, these attached in fresh dishes but were unable to proliferate and indicated cell cycle exit. This corresponded to DNA damage with γH2AX staining and comet assays as above.

Figure 3.

Effects of APAP in HuH‐7 cells synchronized in G1/S by HU. Flow cytometry indicating cells in G1/S after ON culture with HU. After release from G1/S arrest, controls resumed cycling. By contrast, APAP led to loss of S or G2/M fractions in HU‐released cells. Y‐axis, cell number; X‐axis, DNA content. P values were calculated by ANOVA. Studies were repeated thrice

3.2. APAP toxicity affected cell cycle regulators in HuH‐7 cells

In response to APAP toxicity, expression of multiple regulators altered, which was associated with dysregulations in ATM signalling. The relationship between ATM and oxidative stress had been established by others previously.28 The distribution of cell cycle regulators according to IPA indicated that besides ATM signalling, disruptions mapped along G1/S checkpoint, entry into S and G2/M checkpoint (Figure 4, Table 1). These cell cycle proteins included TP53, which was significant, because DDR, including ɣH2AX and MRN complex activations,15 involves checkpoint regulation through CDKN1A (p21), until DNA damage may be repaired by designated proteins, eg, RAD51, etc., or because of irreparable damage cell death decisions may be made, eg, by ABL1, TP73, etc. Mapping of ATM pathways by IPA visualized these mechanisms and was helpful in understanding molecular interactions (Figure S1A). The consequences of ATM signalling on cell cycle proteins seemed to be profound. For instance, decreased CCNE1expression would lead to G/S checkpoint activation since this normally interacts with CCNA and CDK2 to promote cell cycle entry.29, 30 These interactions would be repressed by p21, which enforces cell cycle exit under appropriate DNA damage settings,31 as was observed in HuH‐7 cells surviving APAP toxicity. Moreover, TP53 and p21 regulate E2F‐mediated transcription through Rb, which was also the case in APAP‐treated cells (Figure S1B). Although HuH‐7 cells were primed for entering S with greater expression of CCND1, CDK2 and other proteins, cell cycle suppressors, eg, p21, will again have repressed G1/S progression (Figure S1C). Additional G2 cyclins after APAP toxicity were involved, eg, p21 likely repressed CDK1 and CCNB1 complexing, which is critical to this process (Figure S1D). The effects of ATM signalling were also noted in G2/M checkpoint (Figure S1E). Overall, this mapping of cell cycle regulators was in agreement with APAP‐induced DNA damage and loss of cycling in HuH‐7 cells.

Figure 4.

Overall alterations in cell cycle protein expression in HuH‐7 cells. Groupings of proteins expressed differentially after APAP vs controls in ON culture. The number of proteins with ≥ 2‐fold difference is next to individual bars. Cell cycle categories and ATM signalling were according to IPA. Total number of proteins in canonical pathways in IPA is given. Proteins were listed more than once because of their roles in multiple cell cycle stages. Data were from duplicate arrays with multiple spots per protein in each array

Table 1.

Differentially expressed cell cycle proteins in APAP‐treated vs control untreated HuH‐7 cellsa

ATM signalling		G1/S checkpoint		S entry		G2/M checkpoint
Identity	Expression level	Identity	Expression level	Identity	Expression level	Identity	Expression level
ABL1	4.3	ABL1	4.3	CCND1	2.5	14‐3‐3	18.3
ATM	2.1	ATM	2.1	CCNE1	−3	ABL1	4.3
CCNB1	1.3	CCND1	2.5	CCNE2	4.8	ATM	2.1
CDC25C	16.9	CCND3	3.9	CDK1	1.4	CCNB1	1.3
CDK1	1.4	CCNE1	−3	CDK2	11	CDC25C	16.9
CDK2	11	CCNE2	4.8	CDKN1A	13.3	CDK1	1.4
CDKN1A	13.3	CDK2	11	CDKN1B	6.6	CDK7	38.7
CHEK1	−3.7	CDKN1A	13.3	E2F1	−2.5	CDKN1A	13.3
RAD51	9.5	CDKN1B	6.6	RB1	10.4	CHEK1	−3.7
TP53	2.1	E2F1	−2.4
TP73	1.9	GSK3B	1
		RB1	10.4
		TP53	2.1

^aFold differences vs healthy controls. Lists contain all relevant proteins from IPA networks. Protein expression levels in each pathway differed significantly from controls with P < .001, Mann‐Whitney rank sum tests.

3.3. APAP toxicity in primary human hepatocytes

In human hepatocytes, IC50 of APAP was 20 mM with MTT assays (Figure 5A). This cytotoxicity was associated with significant mitochondrial injury and lower MMP (Figure 5B). The JC‐1 dye multimer/monomer ratios decreased to 0.4 ± 0.1 vs 0.7 ± 0.02 in controls, respectively, P < .05. APAP produced DNA double‐strand breaks with γH2AX in 9.7 ± 0.9 cells per hpf vs 0.4 ± 0.2 cells per hpf in controls, respectively, P < .05 (Figure 5C). To determine whether cell cycling states were altered simultaneously, we examined ploidy class distributions. Due to polyploidy in hepatocytes, individual nuclei may harbour multiple genomes, ie, 2, 4 and 8 genomes in diploid (2N), tetraploid (4N) and octaploid (8N) cells, respectively.32 Under suitable conditions, healthy polyploid cells can replicate although daughter cells may exhibit aneuploid DNA content depending on the extent of asymmetric nuclear divisions.33 However, DNA damage attenuates replication of polyploid hepatocytes, both in vitro and in vivo.32, 34, 35 After ON culture, control primary human hepatocytes (n = 956) were distributed along 2N (22.3%), 4N (48.5%) and 8N+ (29.2%) ploidy classes. In each ploidy class, most hepatocytes had euploid DNA content, which indicated non‐cycling status (Figure 5D). However, in APAP‐treated hepatocytes (n = 300), 2N ploidy class increased to 43.8%, whereas 4N and 8N+ classes decreased, 39.2% and 16.4%, respectively, P < .05 vs controls, ANOVA. The decreases in 4N and 8N+ classes indicated polyploid cells were more susceptible to APAP toxicity. Remarkably, in APAP‐treated cells, aneuploid populations accumulated (Figure 5D), indicating impaired return to euploid DNA states.

Figure 5.

Cytotoxicity of APAP in primary human hepatocytes. (A) Effect of 20 mM APAP on cell viability after ON culture. (B) Oxidative stress after APAP decreased MMP with fewer JC‐1 aggregates (red) vs monomers (green) in ON culture. (C) γH2AX staining (red) for DNA double‐strand breaks in APAP‐treated cells in ON culture. Nuclei were counterstained by DAPI (blue). Original magnifications, x200. (D) Distributions of 2N, 4N and 8N+ nuclei after ON culture with or without APAP. Untreated controls contained euploid DNA whereas aneuploid states were noted after APAP in all ploidy classes. This indicated APAP‐treated cells failed to complete cell division. P values vs controls, ANOVA (A‐C) or Wilcoxon rank sum test (D)

3.4. Cell cycle disruptions were recapitulated in human livers with APAP‐induced ALF

In liver explants, histology verified extensive necrosis and other characteristic features of ALF, eg, steatosis and ductular hyperplasia, whereas healthy donor livers were morphologically normal (Figure 6A). Remarkably, in ALF hepatocytes showed extensive γH2AX staining, which was similar to HuH‐7 cells and primary hepatocytes with APAP toxicity (Figure 6B). Also, p21 was expressed widely in APAP‐induced ALF, indicating downstream effect of ATM signalling (Figure 6C). In healthy livers, we found 0‐1 cells per hpf with Ki‐67‐a marker of all cell cycle stages other than G0/G1.36 In ALF, 2‐10 Ki‐67+ cells per hpf were present, indicating G1, S or G2/M stages. These Ki67+ hepatocytes were in periportal and also mid‐zonal areas of liver lobules where γH2AX and p21 were well‐expressed to suggest DNA damage‐related impairments in cycling.

Figure 6.

ATM signalling was activated in APAP‐induced ALF. (A) H&E staining of liver indicating normal morphology in healthy controls and extensive necrosis in representative sample of explanted liver from ALF. (B) γH2AX staining (red colour) was rare in healthy livers, whereas it was extensive in ALF. (C) Expression of p21 (red) was rare in healthy livers but was widespread in ALF. (D) Ki‐67 expression (red, nuclei) was rare in healthy livers and was more prevalent in ALF. Original magnifications x200; nuclei stained by DAPI (blue). Data were reproduced in independent analyses

Cell cycle protein expression in ALF identified alterations in ATM signalling and G1/S checkpoint, entry into S and G2/M checkpoint (Figure 7A). The up‐ or down‐directionality of cell cycle protein expression in HuH‐7 cells and ALF livers reached consensus in 65% cases, P = .03, χ² with Fisher's exact test. Comparison of hepatic ploidy classes in healthy liver (n = 4014) and APAP‐induced ALF (n = 4514) indicated 2N (17.3%) were outnumbered by 4N (72.2%) or 8N+ (10.5%) classes. This 4N preponderance resembled hepatic ploidy distributions in rats.34 In ALF samples, 2N fraction declined, 4N was similar, and 8N+ increased, 11.0%, 72% and 16.8%, respectively, P < .05 vs controls, ANOVA. The decline in 2N fraction and increase in higher ploidy classes were also similar to restrictions in cell cycling related to PH‐induced DNA damage in rats.34 Notably, in ALF livers aneuploid populations were more prevalent in each ploidy class vs healthy controls, P < .001 (Figure 7B), and indicated cell cycling impairments. The differences in ploidy class distributions in cultured hepatocytes and in explanted livers were likely related to the evolution of APAP‐induced injury over longer periods in the latter condition.

Figure 7.

Cell cycle proteins in human APAP‐induced ALF. (A) Grouping of proteins expressed differentially in ALF vs healthy livers according to IPA. The number of proteins expressed differentially with ≥ 2‐fold differences is indicated next to individual bars. Data were summarized from duplicate arrays. (B) Ploidy distributions of 2N, 4N and 8N+ hepatocyte nuclei in liver from healthy controls and APAP‐induced ALF. Compared with ploidy classes in controls, more hepatocytes after APAP toxicity were aneuploid. P vs controls, Wilcoxon rank sum test

The ATM signalling activation in ALF samples recapitulated that in HuH‐7 cells, except ATM itself was downregulated (Table 2). The decrease in ATM expression was likely due to longer duration from initial liver injury in people as compared with HuH‐7 cells. This was reflected in less DNA damage repair activity, eg, lower RAD51 expression. More CHEK1 protein will have transduced additional signals from ATM‐related protein (ATR) after DNA breaks (Figure S2A).

Table 2.

Differentially expressed cell cycle proteins in APAP‐induced ALF vs control healthy liversa

ATM signalling		G1/S checkpoint		S entry		G2/M checkpoint
Identity	Expression level	Identity	Expression level	Identity	Expression level	Identity	Expression level
ABL1	3.2	ABL1	3.2	CCND1	−2	14‐3‐3	2.2
ATM	−10	ATM	−10	CCNE1	−1.4	ABL1	3.2
CCNB1	−1.5	CCND1	−2	CCNE2	4.6	ATM	−10
CDC25C	−2.4	CCND3	−3.8	CDK1	1.5	CCNB1	−1.5
CDK1	1.5	CCNE1	−1.4	CDK2	1.4	CDC25C	−2.4
CDK2	1.4	CCNE2	4.6	CDKN1A	3.5	CDK1	1.5
CDKN1A	3.5	CDK2	1.4	CDKN1B	2.3	CDK7	3.8
CHEK1	2.2	CDKN1A	3.5	E2F1	6.5	CDKN1A	3.5
RAD51	−5.5	CDKN1B	2.3	RB1	3.9	CHEK1	2.2
TP53	1.6	E2F1	6.5
TP73	5.7	GSK3B	ND
		RB1	3.9
		TP53	1.6

^aFold differences vs healthy controls. Lists contain all relevant proteins from IPA networks. Protein expression levels in each pathway differed significantly from controls with P < .001, Mann‐Whitney rank sum tests.

Mapping of G/S checkpoint confirmed roles of TP53 and p21 in human livers (Figure S2B). Also, CCNE1 was slightly downregulated to indicate possible difficulties in CDK2 complexing, which too was generally similar to HuH‐7 cells. Moreover, CCND1 and CCND3 were expressed less, which too was significant, since this would have impaired entry into S. However, E2F1 was expressed well, which differed from HuH‐7 cells, although the net outcome on LR will still have been negative due to downstream effects on entry into S of TP53 and p21 (Figure S2C).

The G2/M DNA damage checkpoint was affected in human ALF, including decreased CCNB1 expression with likely loss of CDK1 complexing for cell cycle progression. Moreover, CHEK1 would have inhibited other regulators, eg, CDC25C (Figure S2D). Among differences in ALF from HuH‐7 cells were expression of cyclins (Figure S2E), especially CCND1 and CCND2, which increased in HuH‐7 cells but decreased in ALF. This suggested that cell cycle checkpoint activation was more stringent in human ALF.

3.5. Reversibility of cell cycle disruptions in HuH‐7 cells

Previously, ATM‐dependent DDR produced by drug toxicity was restored in HuH‐7 cells by growth factors, such as GCSF, VEGF and others.15 This was of interest because downregulation of ATM promoter transcription was reversed by GMCSF, another haematopoietin,37 whereas GCSF had therapeutic value for ALF in people.38 To determine whether such ATM promoter regulatory mechanism could protect from DNA damage and reverse cell cycle lesions, we performed assays in HuH‐7 cells. Culture of cells with APAP plus GCSF decreased mitochondrial injury in JC‐1 dye assays, along with cytoprotection in MTT assays (Figure 8A,B). These gains were associated with less γH2AX and comet formation (not shown). Flow cytometric profiles indicated GCSF‐treated HuH‐7 cells were no longer depleted in S or G2/M and cycling continued despite ongoing exposure to APAP (Figure 8C,D).

Figure 8.

Effects of GCSF on APAP toxicity in HuH‐7 cells. (A) GCSF restored MMP in APAP‐treated cells with JC‐1 assays, which was obvious after 4 h and ON culture. (B) Cell viability improved after culture with GCSF. (C‐D). Flow cytometry indicating GCSF protected S and G2/M fractions after 4 h with maintenance of cycling similar to healthy controls. Y‐axis, cell number; X‐axis, DNA content. P values were calculated by ANOVA. Studies were repeated thrice

Moreover, cell cycle protein expression was restored significantly in HuH‐7 cells cultured with APAP and GCSF vs APAP alone (Table 3). Restoration of ATM signalling was indicated by normalization of ATM and lowered p21 and TP73 expression, while CHEK1 and RAD51 were expressed more, which should have contributed in DNA damage repair. The G1/S checkpoint was beneficially regulated by normalization of ATM and lowered p21 expression, while expression of the cyclins, CCNE1, CCND1 and CCND3, as well as of E2F1, increased. Similarly, S entry will have been advanced by greater expression of CCND1, CCNE1 and E2F1 and decreased p21 expression. Finally, less expression of 14‐3‐3, ATM and p21 and greater expression of CHEK1 will have removed G2/M checkpoint restrictions. As these effects of GCSF were reflected in protection of mitochondria and cells in S and G2/M, along with continued cycling, this indicated APAP‐induced cell cycle lesions had been reversed.

Table 3.

Cell cycle protein expression in APAP‐treated vs APAP plus GCSF‐treated HuH‐7 cellsa

ATM signalling		G1/S checkpoint		S entry		G2/M checkpoint
Identity	Expression level	Identity	Expression level	Identity	Expression level	Identity	Expression level
ABL1	0.8	ABL1	0.8	CCND1	4.9	14‐3‐3	0.2
ATM	0.5	ATM	0.5	CCNE1	4.2	ABL1	0.8
CCNB1	0.7	CCND1	4.9	CCNE2	1.0	ATM	0.5
CDC25C	1.0	CCND3	2.6	CDK1	0.7	CCNB1	0.7
CDK1	0.7	CCNE1	4.2	CDK2	0.6	CDC25C	1.0
CDK2	0.6	CCNE2	1.0	CDKN1A	0.3	CDK1	0.7
CDKN1A	0.3	CDK2	0.6	CDKN1B	0.8	CDK7	1.0
CHEK1	4.4	CDKN1A	0.3	E2F1	5.8	CDKN1A	0.3
RAD51	2.0	CDKN1B	0.8	RB1	0.7	CHEK1	4.4
TP53	2.0	E2F1	5.8
TP73	0.8	GSK3B	0.7
		RB1	0.7
		TP53	2.0

^aLevel of 1.0 equals protein expression in cells with APAP alone. Levels > 1 and < 1 indicate greater or lower expression in APAP plus GCSF‐treated cells, respectively. Several proteins transducing ATM signalling and DNA damage repair (CHEK1, RAD51) or progression through G1/S checkpoint, S entry (CCND1, CCND3, CCNE1, E2F1) and G2/M checkpoint (CHEK1) were expressed more, P < .003, whereas others imposing restrictions in these stages (ATM, CDKN1A, TP73, 14‐3‐3) were expressed less, P = .02, Mann‐Whitney rank sum tests.

4. DISCUSSION

This study revealed that significant cell cycle lesions in APAP toxicity accounted for failure of LR. These effects of APAP toxicity in the liver had not previously been defined. The HuH‐7 model recapitulated mitochondrial dysfunction, DNA damage and loss of viability, which characterized aspects of APAP toxicity in people previously.4, 5, 8 Various cell lines have been used for APAP toxicity studies, although outcomes may depend in part on gene expression repertoire of cells and also the duration and amount of drug exposure.26, 39 In our studies, IC50 amounts of APAP reproduced acute toxicity in HuH‐7 cells. These effects were verified in primary human hepatocytes, which have been proposed as a gold standard for cell culture studies.40

The spotted protein array was convenient for simultaneous study of multiple proteins, which will otherwise have necessitated cumbersome methods, eg, western blots. In selected instances, we verified results by western blots or immunostaining in HuH‐7 cell samples, eg, expression of ATM, γH2AX and Chek2 phosphoproteins (not shown). In other instances, staining of cells or tissues for γH2AX and p21, confirmed protein expression patterns identified by arrays. The internal consistency of protein expression data was verified by other results. For instance, flow cytometry reflected the consequences of cell cycle lesions, as HuH‐7 cells in S or G2/M were lost rapidly after APAP toxicity. Also, exit from cell cycle with G1/S checkpoint lesions was apparent since cells surviving APAP toxicity were largely in G0/G1. Similarly, absence of cycling in human ALF with DNA damage, p21 and Ki‐67 expression confirmed these findings. Since our focus was on defining mechanisms in failure of LR and not the kinetics of the onset of LR, it was appropriate to examine cells or tissues for these parameters after APAP toxicity and its effects on cell cycling were established.

Mitochondrial injury has assumed significant pathophysiological roles in APAP toxicity.4, 8 We found MMP was markedly impaired after APAP toxicity in both HuH‐7 cells and primary human hepatocytes. This was associated with DNA damage. Previously, mitochondrial DNA depletion due to DGUOK, POLG, MPV17 or TRMU gene deficiencies caused hepatic damage and ALF in early life.19, 20, 21 Insufficiency variants in these genes increased susceptibility to ALF due to other causes.22 Moreover, mitochondrial DNA depletion was observed in people receiving liver transplantation for ALF.41 Hepatic injury with natural toxins and drugs may also deplete mitochondrial DNA, eg, Wilson's disease (copper), hereditary tyrosinaemia (metabolic), cholestasis (bile salts) and valproate or isoniazid.22

The role of ATM in mitochondrial viability has critical biological significance. Reactive oxygen species and oxidative stress may directly activate ATM and cell survival signalling, eg, p38 mitogen‐activated protein kinase (MAPK), and proliferation inhibitors, eg, TP53 and others.28 Previously, activation of p38 MAPK and other cell cycle regulators after oxidative hepatic DNA damage with ischaemia‐reperfusion constituted a protective response, but at the cost of impaired LR capacity.42 With overwhelming cell injury, ATM itself may undergo oxidative damage.43 This may explain lower ATM protein levels in human ALF and should be significant since loss of ATM activity impaired DDR and restricted cycling.13, 15, 16, 18 The profound consequences of ATM deficiency on cell survival are established, eg, in ataxia telangiectasia,13 and impaired LR following PH in ATM knockout mice.17 Also, ATM phosphorylates the cytoprotective transcription factor, STAT3, which benefits mitochondrial function and clearance.24, 25

In HuH‐7 cells, APAP toxicity was rapid with γH2AX expression, comets and clearance of cells already in S and G2/M by 4 hours. This susceptibility of actively cycling cells was in agreement with greater damage in replication‐ready DNA.44, 45 Although healthy hepatocytes in adult liver are thought to be in G0/G1, most are polyploid, with attenuations in replication capacity after DNA damage.32, 34, 35 For such polyploid cells in the liver, attempted synchronous entry into G1/S or post‐G1/S during LR could be catastrophic in view of their susceptibility for additional DNA damage and rapid clearance in case of APAP toxicity. Remarkably, studies here showed hepatocytes accumulated in aneuploid states after APAP toxicity in vitro as well as in vivo, indicating cell cycle restrictions.

On the other hand, cell cycle exit following DNA damage in residual hepatocytes, as indicated by flow cytometric and other results in HuH‐7 cells, as well as aneuploid divisions in ALF, would pose barriers to LR. The overarching regulation by TP53 and p21 of G1/S checkpoint related to CCNE/CDK2 and Rb/E2F interactions, S phase entry related to CCND1 and CCNE and partners, and G2/M checkpoint involving CCNB and partners, allowed insights into restriction of LR in ALF. These regulatory interactions were important because CCNE/CDK2 complexing normally advances cell cycle,29, 30 whereas p21 contributes in cell cycle exit by inhibiting these interactions.31 Similarly, p21, in conjunction with CCNE and Rb, inhibits entry into S.46

Importantly, mitosis is arrested without distribution of healthy mitochondria to daughter cells.23 Recently, the critical role of Drp‐1 or DNM1L and FIS1 in mitochondrial fusion and fission and their regulation by cell cycle genes came into focus. Maintenance of Drp1 requires CDK1 and CCNB complex, whereas CCNE is required for Drp‐1 or DNM1L‐dependent hyperfusion of mitochondria during G1/S transition.47 This is followed by mitochondrial fragmentation by Drp‐1 and FIS1 in S and G2/M under control of CCNB/CDK1 complex. It should be noteworthy that low MMP and decreased cellular energy levels lead to arrest in G1/S.47, 48 Remarkably, loss of Drp‐1 led to mitochondrial hyperfusion with activation of additional ATM‐dependent G2/M arrest and aneuploidy.49

These cell proliferation controls are significant for APAP toxicity because compared with the knowledge of cell cycle regulation in cancers,50 this is less well understood in hepatic drug toxicity. Previously, impairment of PH‐induced LR by mutagenic injury with 2‐AAF, rapidly increased TP53 and p21 expression with lower CCNE and Rb expression preceding G1/S checkpoint.12 Also, simultaneous S and G2/M checkpoint lesions were noted in arsenic toxicity with accumulations in G2/M before apoptosis.51 Other connections were made between ATM and inflammatory cytokines, eg, TNF‐IL6‐NFkB.52 This should be relevant for LR in ALF, eg, inhibition of TNF‐α signalling helped control inflammation and improved hepatocyte survival and proliferation after transplantation.53 When APAP was combined with other toxins, eg, alcohol, liver injury worsened.54

The reversibility of cell cycle lesions by GCSF may provide avenues to overcome failure of LR (Figure 9). Previously, GCSF, VEGF and other growth factors, improved ATM signalling and decreased injury in HuH‐7 cells.15 Remarkably, GCSF decreased mitochondrial injury and DNA damage, along with restoration of ATM‐dependent pathways and cell cycle regulators to advance G1/S, entry into S and G2/M traverse. Since ATM is additionally implicated in stem cell renewal,55, 56 recruitment of resident stem/progenitor cells during LR could be another avenue for overcoming cell cycle deficits. In prior studies, GCSF promoted expansion of putative liver stem cells as well as of other cell populations during PH‐induced liver remodelling.57, 58

Figure 9.

Simplified model for restoring LR after APAP hepatotoxicity. (A) Indicates key problems identified in this study with ATM pathway‐related onset of hepatic DNA damage and multiple cell cycle lesions. (B) Critical points for therapy will concern restoration of ATM signalling with normalization of CHEK2 and DDR pathways. Simultaneously, normalization of cell cycle checkpoints, particularly TP53 and p21 cascade, and cell cycle advancement through cyclins, their partners and other contributors, will be helpful

CONFLICT OF INTEREST

The authors declare no competing financial interests or conflicts in relation to the work described.

AUTHOR CONTRIBUTIONS

PV, YS and PG performed studies and interpreted data; SG designed study, analysed and interpreted data; PV and SG drafted manuscript; all authors reviewed and approved manuscript.

Supporting information

 

P Viswanathan, Y Sharma, P Gupta, S. Gupta Replicative stress and alterations in cell cycle checkpoint controls following acetaminophen hepatotoxicity restrict liver regeneration. Cell Prolif 2018; 51:e12445 10.1111/cpr.12445.