Resveralogues protect HepG2 cells against cellular senescence induced by hepatotoxic metabolites

.


Introduction
The mammalian liver performs hundreds of important functions including the manufacture of bile, plasma protein synthesis, xenobiotic metabolism and the conversion of ammonia to urea.Liver dysfunction increases markedly with ageing and can occur both through primary liver diseases and secondary causes such as metabolic syndrome.For example, non-alcoholic fatty liver disease increases in both frequency and severity in the over 65 age-groups (Bertolotti et al., 2014) and similar patterns are observed for both alcoholic and autoimmune liver disease (Tajiri and Shimizu 2013).The high prevalence of polypharmacy in the elderly also increases the need for effective xenobiotic metabolism.
A general feature of liver dysfunction and impaired liver clearance is the accumulation of toxic metabolites such as unconjugated bilirubin and ammonia in the plasma.These in turn produce life-threatening effects in other organs, such as neurotoxicity and hepatic encephalopathy, contributing to the high mortality associated with liver failure (Bleibel et al., 2012).Although the preferred treatment for liver failure is transplantation there is a serious shortage of organ donors and both increasing donor and patient age independently predict poor transplant outcomes (Dayoub et al., 2018;Minich et al., 2022).Thus, unless other solutions can be found, the scale of this problem is likely to increase as the global population ages.An alternative approach to transplantation is the development of bioartificial liver systems (BALS).These are artificial organs intended either to act as an emergency extracorporeal bridge for patients until a suitable donor organ is available or to support endogenous organ function until the patient's own liver recovers.To achieve this BALS must replace the primary metabolic activities of the liver for which the system in turn relies on 400-1000 g of either porcine or human hepatocytes (in the latter case typically HepG2 sub-clones) to provide the necessary biotransformation and synthetic functions (Struecker et al., 2014).In a typical BAL design the patient's blood or plasma is filtered through a device in which these hepatocytes are housed, such as a hollow fiber bioreactor (in which the cells and plasma are on opposite sides of the semipermeable membranes), a multilayer membrane or a scaffold-based system (in which cells may be separated from, or in direct contact with, patient plasma) and then returned.Hollow fiber-based BAL systems have been extensively tested in clinical trials.However, evidence for improved patient survival is lacking and, for unknown reasons, contact with the plasma of individuals with liver failure severely reduced ammonia clearance, urea synthesis and xenobiotic metabolism even in autologous primary hepatocytes (Abrahamse et al., 2002).Although overall cell viability is unaffected this loss of function completely compromises device utility and represents an important barrier to the development of working BAL systems.
This combination of alterations in phenotype with unaffected viability in response to physiological stressors is highly reminiscent of cellular senescence (Faragher et al., 2017).A wide range of stressors induce the senescent state (including sustained cell turnover, oncogene activation and ER stress) which acts an in vivo barrier to carcinogenesis early in life (the altered phenotype and particularly its sterile inflammatory component facilitates clearance of senescent cells by the natural killer (NK) component of the immune system) causing ageing later in the life course as clearance fails (Bai et al., 2022;Boccardi et al., 2024).Transgenic models in which senescent cells can be cleared artificially (e. g. by senescent cell-specific expression of Herpes Simplex Virus thymidine kinase followed by exposure to ganciclovir in vivo) provide interventional evidence that such cells are primary causes of both morbidity and mortality.Whilst these discoveries have provided a powerful impetus for the identification of compounds which selectively kill certain types of senescent cells we have recently demonstrated that it is also possible to reverse replicative senescence in multiple primary human cell types through the use of a library of resveralogues -novel compounds based on the stilbenoid resveratrol (Latorre et al., 2017;Birar et al., 2020).This rescue is associated with transient lengthening of telomeres and occurs independently of either the blockade of proinflammatory cytokine release or the interaction with resveratrol's canonical target sirtuin 1 (SIRT1).Activation of this NAD-dependent deacetylase enhances autophagy and antioxidant defense mechanisms indirectly via FOXO3a, p53 and PGC-1α (Singh et al., 2018) and compounds from our library which are SIRT1 activators show enhanced senescence rescue effects over closely related resveralogues which are not (Birar et al., 2020).
Senescence in response to DNA damaging agents has been reported previously in HepG2 cells (Aravinthan et al., 2014) and been shown to be associated with alterations in gene expression although the primary synthetic functions of hepatocytes, such as albumin and urea production, were not directly investigated.Based on this we hypothesized that one or more components of the hepatotoxic mixture of metabolites and cytokines that accumulate in plasma during liver failure induces senescence which, in turn, disrupts features of the HepG2 phenotype central to liver detoxification functions.
Accordingly, for the first time, we have measured the effects of physiologically relevant concentrations of hepatotoxic metabolites on HepG2 cells grown both in standard two-dimensional cultures and in novel p(HEMA)-alginate cryogel systems.We have previously demonstrated that these hepatocyte loaded cryogels have significant potential as bioreactor scaffolds for BALS due to their simultaneous reduction of protein fouling and enhancement of hepatocyte growth, albumin and urea synthesis (Bonalumi et al., 2021).Simultaneously the potential for SIRT1 activating and non-activating resveralogues to protect HepG2 populations from hepatotoxic metabolites were evaluated.

Synthesis of poly (HEMA-co-MBA) cryogels
As previously described (Bonalumi et al., 2021) cryogels made of 2-hydroxyethyl methacrylate(HEMA) (Sigma UK) and cross-linked with N, N′-methylenebis (acrylamide) (MBA) (Sigma UK) were synthesised for use as three-dimensional BAL cell scaffolds.Briefly, 0.891 mL of monomer HEMA and 0.266 g of cross-linker MBA was dissolved in 20 mL of deionised (DI) water in a conical flask and the solution was degassed with nitrogen gas or a vacuum pump for 10 minutes to remove the soluble oxygen.The cryogel was produced by free-radical polymerisation catalyzed by N, N, N′, N′-tetramethylethylenediamine (TEMED) (Sigma, UK) and ammonium persulfate (APS) (Sigma, UK).After adding 20 µL of TEMED (first initiator) the solution was cooled in an ice bath for 15 minutes.20 mg of APS (second initiator) was added to the solution and the reaction mixture was stirred for 30 seconds.A volume of 2 mL was immediately pipetted into a 9 mm inner diameter glass tube.The solutions were frozen in an ethanol cryobath at − 12 • C for 18 hours.The cryogel columns were thawed at room temperature, washed with 500 mL of DI water and stored at 2-8 C for further use.

Post-synthesis modification of HEMA-MBA cryogels with alginate.
HEMA-MBA cryogels were incubated with agitation for 2 hours in an aqueous solution of 0.1 M glutaraldehyde (Sigma, UK) and 0.1 M HCl (Sigma, UK).After washing the cryogels three times in deionised water, they were incubated overnight with agitation in a 1% (w/v) solution of alginate w/v (Sigma, UK).Finally, the cryogels were washed 3 times with water and stored in deionised water at 2-8 C for further use.

Cell line and culture conditions
HepG2 (human liver hepatocellular carcinoma) cells were purchased from the American type culture collection (ATCC) and cultured in standard tissue culture plastic vessels containing Eagle's Minimum Essential Medium (MEM) (Thermo Fisher, UK) supplemented with 10% (v/v) FBS, 1% NEAA (v/v) and antibiotics in a humidified incubator with 5% CO 2 at 37 • C. The media was changed every alternate day.Cells were passaged at 70% confluency by standard Trypsin-EDTA (Thermo Fisher, UK) dispersion and seeded at 6500 cells cm-2 .

Induction of senescence
HepG2 cells were cultured as described above, seeded on 2-dimensional (13 mm glass coverslips) and 3-dimensional (alginate cryogel) matrices at 1×10 5 cells mL − 1 .Control cultures were incubated with standard media alone whilst test populations were treated with basal medium supplemented with the standard mixture of hepatotoxic metabolites and/or cytokines described below for 6 hours.After this the cultures were washed with PBS once and allowed to recover in complete culture medium overnight.The toxic metabolites used, and their concentrations, are shown in Table 1.

MTT assay
Cell viability was measured by thiazolyl blue tetrazolium bromide (MTT) (Abcam, UK) assay comparing 2-D and 3-D cryogel surfaces.A standard curve was used to establish a linear relationship between cell concentration and MTT absorbance values using serial dilution in the cell concentration range from 5 ×105 cells to 0. To measure cell viability HepG2 cell was seeded into the wells of 96 well plates (2-D surfaces) and onto alginate cryogels at 5×105 cells/mL concentration in duplicate.After 24 h of incubation, 10 µL of 10% MTT was added to each well and incubated at 37 • C (5% CO2) for 2 h until purple-colored formazan product developed.At the end of the incubation, the medium was removed from each well and 100 µL of dimethyl sulfoxide (DMSO) (Thermofisher, UK) was added to each well to lyse the cells.Following 5 minutes incubation in DMSO the absorbance was measured by spectrophotometry at a wavelength of 570 nm.
In order to measure the cell viability on cryogels, after the extract incubation time, the discs were moved to a new 24 well-plate and a 500 µL solution of 10% MTT (5 mg/mL) in phenol red-free medium was added to each well followed by a further 4 h incubation at 37 C in a humidified incubator with 5% CO2.At the end of the incubation, the medium was replaced by 500 µL of DMSO and 100 μL aliquots were transferred to a 96 well plate.The absorbance was read at 570 nm.

Live and dead assay
A live/dead cell viability assay (Thermo Fisher, UK) was performed, seeding HepG2 cells at a concentration of 5×105 cells/mL on HEMA-MBA-alginate cryogels.The cell-seeded cryogels were incubated at 37 C in a humidified incubator with 5% CO2 for 6 days.The medium was changed every 2 days.Before the assay, the polymers were washed with PBS to remove or dilute serum esterase activity generally present in serum-supplemented growth media.A 100 µL of live and dead working solution including calcein AM and ethidium homodimer-1 was added to each of the cryogel discs followed by visualisation using confocal microscopy.Calcein-AM positive cells were detected at 494/517 nm and ethidium homodimer positive cells were detected at 528/617 nm excitation/emission wavelength.

Hepatotoxin protection studies
Cultures of HepG2 cells grown under either two-or threedimensional conditions as described in 2.4 above were incubated in media containing resveralogues (resveratrol, dihydroresveratrol, V29, V31 and V34 -synthesis described in Birar et al., 2020) at 2-5μM for 24 h.After this, the medium was changed to fresh resveralogue supplemented medium which also contained the standard hepatotoxin and/or cytokine cocktail described above for 6 h.After this challenge, the medium was removed and the cultures incubated in standard growth medium overnight.Control populations were maintained in growth medium only.

Determination of culture growth fraction by EdU incorporation
The Click-iT EdU Alexa Fluor 488 imaging kits (Thermofisher C10337, UK) were used to determine S phase traverse in accordance with manufacturer's protocol.Briefly, HepG2 cells were incubated with 10 μM 5-ethynyl-2′-deoxyuridine (EdU) for 3 h.Then cells were rinsed with medium and fixed using 3.7% (v/v) formaldehyde for 10 minutes.Fixed cells were permeabilised using a solution of 0.5% (v/v) Triton-X-100 in PBS for 20 min then washed twice with PBS and treated with 50 µL of the Click-iT reaction cocktail for 30 min.The cells were then washed once with PBS and counterstained with 1 μg mL − 1 Hoechst 3342 in PBS for 10 min.Fluorescence was measured using a confocal microscope with excitation/emission filters set at 495/519 nm to detect Alexa Fluor® and at 350/461 nm to detect Hoechst 3342.To determine the proliferating fraction, either a minimum of 500 total nuclei were counted in random fields on each of three matrices.

Immunocytochemical detection of Ki67 and p53
HepG2 cells were seeded at 1×10 5 cells mL − 1 in a 6-well plate containing autoclave sterilised coverslips.Cells on coverslips were washed twice in 1 mL of PBS and fixed with 3.7% formaldehyde solution v/v (Sigma, UK) for 10 minutes at room temperature, washed twice with wash buffer (3% BSA in PBS, w/v) and then permeabilised with 0.5% Triton® X-100 (Triton, Sigma) v/v in PBS.After three PBS washes, the cells were incubated with 1 µg/mL Ki67 monoclonal antibody (Thermofisher, UK) and 1 µg/mL of anti-p53 mouse monoclonal antibody (Anti-P53 Antibody, Clone BP53-12, Sigma, UK) at 4 • C overnight.Then, cells were washed with wash buffer 10 times and incubated with 1 µg/ mL Goat anti-Mouse IgG (H+L), Superclonal™ recombinant secondary antibody, Alexa Fluor 488 (Thermofisher, UK) at room temperature for 4 h.The nuclei stained with DAPI and visualised using a confocal microscope and filter were set for Alexa Fluor 488 with an excitation/ emission wavelength of 495/519 nm and excitation/emission wavelength of 405/488 nm to detect DAPI stained nuclei.The percentage of ki67 and p53-positive cells were quantified by counting the number of Ki67 and p53 positive cells (green stained) as a proportion of total blue stained, DAPI positive cells.Cells traversing the cell cycle showed a distinctive pattern of nuclear staining.To determine the proliferating fraction, a minimum of 500 nuclei were counted in randomly selected fields.Experiments were repeated in triplicate.

SA-β-gal staining
The senescence-associated beta-galactosidase activity in HepG2 cells treated with etoposide and liver toxins, with or without resveratrol and resveralogues, was measured using the Senescence β-Galactosidase Staining Kit (Cell Signalling Technology, USA), following the manufacturer's instructions.Cells were examined via a light microscope, and the percentage of SA-β-gal-positive cells was calculated.

Determination of albumin production by ELISA
For two-dimensional cultures, HepG2 were seeded on a 24-well plate at a final concentration of 5×10 5 cells mL − 1 .For 3 dimensional cultures the cell suspension was seeded on HEMA-MBA and HEMA-MBA-alginate matrices at a final concentration of 5×10 5 per matrix in duplicate and allowed to adhere for 24 h.They were treated with liver toxins for 6 h.The medium then were removed from each well and replaced with standard growth medium only for 48 h.The media was collected from each sample after 48 h and immediately frozen and stored at − 20 • C until analysis.Cell free medium served as a control.
Albumin levels were determined using the SimpleStep ELISA® kit (Abcam UK) in accordance with the manufacturer's protocol.Briefly, 50 µL of each sample and standard solutions were added to an anti-tag antibody pre-coated 96 well plates.Then, 50 µL of an antibody cocktail containing capture and detection antibodies was added to each well and the plate sealed and incubated for 1 hour at room temperature on a plate shaker set to 400 rpm.The wells were then washed 3 times with 350 µL 1X Wash Buffer PT; 100 µL of 3,3′,5,5′-Tetramethylbenzidine (TMB) substrate was added to each well and incubated for 10 minutes in the dark on a plate shaker at 400 rpm. 100 µL of stop solution was then added to each well and the plate was covered and incubated for 1 min in the dark on a plate shaker at 400 rpm.The absorbance was read on a HepG2 cells treated with 10 µM or 20 µM etoposide for 48 hours to induce replicative senescence were used as controls.After treatment, the cultures were washed with PBS and allowed to recover with a complete culture medium for 1 day before use.
N. Heidari et al. plate reader at 450 nm.Albumin concentrations were calculated using a complementary standard curve.

Determination of urea production
In 24 well plate cultures of HepG2 cells identical to those described in 2.8 above urea production was stimulated by treatment with 1 mM ammonium chloride (NH 4 Cl), (Sigma 213330, UK) for 24 h after which the medium was collected and stored at − 20 • C. The urea concentration was then determined using a Quantichrom Urea assay kit (Bioassay Systems DUR-100, US).Briefly, 50 µL medium (blank), 50 µL of 5 mg urea dL − 1 (standard) and 50 µL of all samples were transferred in duplicate into 96 well plates.Then, 200 µL of working reagent comprising a 1:1 mixture of reagent A and reagent B (provided in the kit) was added to each well and incubated at room temperature for 20 minutes.The absorbance was then measured with the plate reader at 520 nm.The urea concentration in each sample was calculated following this equation: OD sample, OD blank and OD standard are the absorbance of sample, blank and standard respectively, n is the dilution factor and [STD] is 5 mg dL − 1 for low urea-containing samples.

Statistical Analysis
Data was expressed as mean +/-standard error of the mean using at least triplicate experimental repeats and duplicate repeats within each experiment.

Real-time reverse transcription polymerase chain reaction (RT-PCR)
HepG2 cells at 1×10 5 cells mL − 1 concentrations were seeded into the wells of a 6-well plate and incubated at 37 C, 5% CO 2 overnight.Cells were treated with liver toxins cocktail for 6 h and control cells were grown in media only.After 6 h incubation, the treatment was washed with twice PBS and trypsinised after 24 and 48 h recovery time in complete medium.Each cell pellet was collected and stored at − 80 C until RNA extraction.At this point cell pellets were washed with 1 mL of PBS and centrifuged at maximum speed for 30 seconds.350 µl of buffer RLT and 350 µl of 70% ethanol (RNAeasy Mini Kit, Qiagen, UK) were added to the samples which were then transferred to the RNA easy Min Elute spin column.DNase and any contaminants were washed with 700 µl of RW1 buffer and 500 µl of buffer RPE respectively.Then, the total RNA was eluted in RNase-free water.Total RNA concentration and purity were measured by the Nanodrop machine (2000/2000 C Spectrophotometers, Thermo Scientific, UK) by the 260/280 nm absorbance ratio.Template RNA was subjected to Quantitect reverse transcription procedure for complementary DNA (cDNA) synthesis.The purified RNA sample was incubated in a gDNA wipeout buffer (Qiagen, UK) at 42 C for 2 minutes to remove contaminating genomic DNA.The cDNA was synthesized by reverse transcription using a master mix (Qiagen, UK) (1 µl of quantiscript reverse transcription, 2 µl of quantiscript RT buffer and 1 µl RT primer mix) according to the manufacturer's protocol.The entire reaction was kept at 42 C for 15 minutes and then reverse transcriptase was inactivated at 95 C for 3 minutes.Gene expression was measured on Rotor-Gene Q (Qiagen, UK) cycler according to the program outlined in the manufacturer's protocol using SYBER green master mix (Qiagen, UK).PCR was done in the condition of denaturation at 95 C for 5 min and combined annealing/extension at 60 C for 10 S for 40 cycles.The primer sequence for real-time PCR is shown in Table 2. Relative gene expression levels were calculated using the 2 -ΔΔCT method.GAPDG was used as an internal reference.

Data and statistical analysis
All data were analysed using Graphpad Prism 9 software.Data values are displayed as the mean ± standard deviation, derived from three different biological replicates.Statistical analyses were performed using one-or two-way ANOVA with Bonferroni 's multiple comparison test.

Etoposide treatment induces cell senescence in HepG2 cells which compromises key liver specific functions
To determine the effects of entry into the senescent state on some of the primary functions of hepatocytes a well characterized method of inducing cellular senescence in HepG2 cells (treatment with the DNA damaging agent etoposide) was initially employed.Population viability, as measured by the MTT assay, after 24 and 48 h was 100%±0.28 and 84%±0.2respectively in control cells, 98.8%±0.29 and 83.7%±0.21 in 10µM-treated HepG2 cells, 71%±0.01 and 70%±0.024 in 20 µM treated cultures.10 µM etoposide had no significant effect on cell viability after 48 h relative to controls (Fig. 1a).Consequently, this concentration and exposure time were used in further experiments.
Following etoposide treatment two dimensional HepG2 cultures labelled with EdU for three hours displayed a highly significant (p˂0.001)reduction in the percentage of positive nuclei from 29±3% in controls to 19.6±2% and 15±2% after 48-hour treatment with 10 µM or 20 µM etoposide (Fig. 1b).A more sensitive assay for the detection of all cycling cells using immunocytochemical visualisation of Ki67 protein revealed that the total cycling fraction of treated HepG2 cultures fell from more than 58% positive nuclei in controls to less than 20% in etoposide treated populations (Fig. 1c).
This decline in the growth fraction was associated with the stabilization of p53 protein in the etoposide-treated cultures (Fig. 1d).48% of cells exposed to this DNA damaging agent displayed positive staining for p53 which was significantly higher (p˂0.0001)than the control group (13.9% positive staining).In parallel with p53 immunostaining the expression level of the p53 gene was assessed using RT-PCR.Fig. 1e shows the comparison of p53 gene expression levels in treated cells versus control.The results indicate a notable upregulation of p53 expression at the message level in etoposide-treated HepG2 cells compared to controls (p<0.05).
Senescence-associated β-galactosidase (SA-β-Gal) activity is commonly used to visualize senescent cells and typically becomes detectable several days after entry into the senescent state.SA-β-Gal at pH 6.0 was measured 7 days after treatment with 10 µM etoposide.As shown in Fig. 1 f, the percentage of SA-β-Gal positive cells increased  significantly compared to non-treated cells (p< 0.01).HepG2 cells were subsequently cultured under standard twodimensional conditions as well as in three dimensional cryogel matrices which are known to enhance the activity of primary hepatocyte functional pathways, including albumin and urea synthesis ( (Bonalumi et al., 2021).As expected, HepG2 populations displayed an increased capacity to produce albumin and urea under these conditions compared to standard two-dimensional cultures (Fig. 1 g and 1 h).However, this was sharply curtailed under any culture conditions if the cells were treated with 10 µM etoposide (p˂0.05 for albumin and p ˂0.001 for urea synthesis).

Exposure to liver toxins induces a senescent state in HepG2 cultures
To determine whether a similar senescent state could be induced by exposure to the metabolic by-products resulting from dysregulated liver function; HepG2 cells were cultured for six hours in media containing a standard cocktail of hepatotoxic compounds and pro-inflammatory cytokines at concentrations typical of those encountered in vivo.Exposure to the toxin/cytokine mixture had no statistically significant effect on HepG2 viability as measured by either MTT or Live/Dead assays (Fig. 2a).The cycling fraction of cells in the population, as measured by EdU label incorporation, was dramatically reduced (from ~76% to ~15%, p< 0.0001).A similar reduction was noted by Ki67 immunostaining (reduction from ~82% to~40%, p<0.001).The relevant data are shown in Figs.2b and 2c respectively.
Upregulation of the activity and/or amount of p53, p21Waf and p16INK4a, are well established early markers of senescence (reviewed in Faragher; 2021).Therefore, their levels were measured by qPCR 24-and 48-hours post-treatment with liver toxins.Fig. 2d shows that p21 and p16 expression levels in the liver toxins-treated group were significantly higher than control group (p˂ 0.05).The level of p53 message increased in the liver toxins-treated group compared to the control group.However, the difference is not statistically significant.This senescent state is marked by an initial upregulation of p53 and p21 at the message level followed by an upregulation of p16.This combination of a reduction in labelling index and an upregulation of cyclin dependent kinase inhibitors in HepG2 populations following liver toxin treatment is consistent cell senescence rather than a cytostatic arrest (e.g. that produced by the double thymidine block technique in cell synchronisation studies).The appearance of a large SA-β-Gal positive population several days after hepatotoxin treatment is further evidence that one or more N. Heidari et al. components of this mixture trigger senescence in HepG2 cells (Fig. 4e).
In order to determine whether the cytokine or hepatotoxic components of the mixture were primarily responsible for the induction of senescence HepG2 cultures were incubated with single cytokines, with hepatotoxic compounds alone or with the standard mixture of both.As can be seen in Fig. 2e, only IL-8 produced a small, but statistically significant, reduction in the growth fraction (p˂0.05).In contrast the cocktail of hepatotoxic metabolites alone reduced the growth fraction from 57 ± 1.4% to 18 ±1% (p˂0.0001) with a further modest reduction if cytokines were included in the mixture (15 ± 0.3%).

Senescence induced by liver toxins compromises key liver specific functions
To determine if this state of Senescence Induced by Liver Toxins (SILT) compromised primary hepatocyte functions as observed with etoposide treatment, the levels of albumin and urea synthesis were determined under both two dimensional and three-dimensional culture conditions (Figs. 3a and 3b).Again, the effects of exposure to realistic in vivo concentrations of hepatotoxic metabolites were dramatic.Albumin synthesis in HepG2 cultures dropped more than 20-fold after six hours of exposure (from 0.47 ± 0.05-0.019±0.04 mg mL − 1 , p˂0.01) whilst urea synthesis was reduced nine-fold (12.08±0.40-1.33±0.80mg dL − 1 , p˂0.001).These effects were largest on HepG2 cell populations grown in three-dimensional matrices, precisely because of the upregulation of activity produced by culture on cryogels.

Resveratrol protects from SILT
To determine whether SILT can be prevented, cultures of HepG2 cells grown on both two-and three-dimensional matrices were preincubated with 2-5 µM resveratrol for 24 h prior to exposure to medium containing the standard cocktail of hepatotoxic metabolites used previously but supplemented with 2-5 µM resveratrol.As in previous experiments, control cultures not treated with resveratrol rapidly entered SILT as measured by EdU incorporation (p<0.001).However, cultures treated with resveratrol were essentially completely protected from SILT after 6 h (Figs.4a-4d) p<0.01 vs liver toxin treated cultures, no statistical difference to untreated controls.Consistent with our earlier work (Birar et al., 2020) a significant increase in the growth fraction was detectable in HepG2 cells treated with resveratrol alone (Supplementary data, Figure S6).Again, detection of the total fraction of cycling cells by immunostaining for pKi67 showed a highly significant reduction in the labelling index on exposure to liver toxins which was completely prevented by supplementation of the medium with resveratrol (23±0.04%positive nuclei in cultures exposed to toxins alone versus 56±0.4% and 58±3.3% for those supplemented with 2-5 µM resveratrol).
A significant elevation in the percentage of SA-β-gal positive cells was observed following exposure to liver toxins, as evidenced by the marked increase when compared to the control cells (***p<0.001,see Fig. 4e).Approximately 37% of hepatotoxin treated cells displayed positive staining for SA-β-gal within three days of the removal of the treatment.However, in cultures treated with 5 µM resveratrol, there was a substantial reduction in the percentage of β-gal positive cells (****p<0.0001),consistent with a protective effect against the impact of SILT (Fig. 4e).

Resveralogue mediated protection from SILT also protects from key hepatocyte functions and is partially independent of SIRT1 activation
To probe the potential mechanism of action by which protection from SILT occurs, HepG2 cells in two-dimensional culture were challenged with our hepatotoxic cocktail after pre-treatment at 5 µM with dihydro resveratrol and a structural series of novel resveralogues V29, V31 and V34 (Birar et al., 2021).These differ both in their endogenous antioxidant capacities and ability to activate SIRT1, a canonical target of resveratrol.As shown in Fig. 5a all the resveralogues were able to protect from SILT (p<0.001) although protection was lower, but still highly significant (p<0.01) with V29 which lacks the capacity to activate SIRT1.
In addition to the observed effects of resveralogues on the growth fraction, a significant reduction in the percentage of SA-β-gal positive cells was observed in cultures treated with V34 and V29, compared to cells exposed to liver toxins (p<0.001).This finding indicates a consistent trend, suggesting that both V34 and V29 exert a protective influence on senescence-induced growth arrest (SILT), akin to the impact observed with resveratrol treatment (Fig. 5b).
To determine whether the resveralogues were equally effective at protecting key liver specific functions three dimensional cultures of HepG2 cells were protected with the same resveralogue series described    N. Heidari et al. above and then challenged with hepatotoxic metabolites for 6 h.All the resveralogues proved equally effective in protecting the capacity of the cultures to produce albumin although V29 was somewhat less effective that the others (p<0.05) in protecting the biosynthetic capacity for urea after toxin challenge (Figs.5c and 5d).

Discussion
Previous studies have shown that HepG2 cells can be induced to enter a senescent state by treatment with DNA damaging agents and that this state of replicative senescence is associated with significant shifts in phenotype, including the upregulation of p53 dependent transcripts, altered oxidative stress responses and dysregulated endocytosis (Nagano et al., 2016).But, perhaps surprisingly, far less attention has been paid to senescence-induced changes to the core functions that hepatocytes normally discharge in vivo.This is possibly because the HepG2 line was originally isolated from a well differentiated hepatocellular carcinoma and the cells are known to show lower albumin and urea synthesis than their primary counterparts.Nonetheless HepG2 cells perform many differentiated hepatic functions and the line is widely used in drug metabolism and hepatotoxicity studies (Donato et al., 2015).Moreover, HepG2 and its various subclones are currently the only realistic source of hepatocyte material for BALS that do not carry the risk of interspecies viral transmission seen, for example, in the use of primary porcine hepatocytes.
Conscious of this, we have demonstrated for the first time that etoposide treatment (a well characterized inducer of senescence in multiple cell types including HepG2) severely compromises both albumin and urea production in HepG2 cells.Crucially we have shown that a physiologically reflective mixture of the hepatotoxins found in the plasma of patients with liver failure also cause albumin and urea production to drop dramatically and induces senescence, as measured by the cessation of proliferation, upregulation of some cyclin dependent kinase inhibitors responsible for initiating and maintaining the senescent state and eventually the accumulation of SA-β-gal positive cells.Senescence occurs over the time course in which patients are typically connected to BALS rendering any hepatocytes within the device essentially useless for plasma detoxification.Several days after hepatotoxin treatment senescent HepG2 cells also overexpress a number of proinflammatory cytokines reminiscent of the well-known senescence associated secretory phenotype or SASP (see supplementary data Figure S2).However, because of the delay between senescent stimulus and SASP production common across many cell types (Faragher, 2021) we do not consider this to be a major issue for BALS except in treatment regimes in which the intent is to reuse cartridges of cells after they have been linked to patients in the bioartificial organ system.
This previously unrecognised SILT effect may prove to be a major, if not the sole, reason for the failure of several highly promising BAL designs.The earlier work of Abrahamse et al. (2002) who demonstrated failure of the AMC-BAL device to protect anhepatic pigs despite bioreactors being populated with their own autologous primary hepatocytes is particularly informative in this regard.The combination of loss of the ability to produce albumin or urea in primary porcine hepatocyte populations with undiminished cell viability reported by these researchers after exposure to the hepatotoxins in the plasma of these animals is strongly reminiscent of SILT.This loss of hepatocyte function, and thence the ability to keep the donor animals alive, clearly does not result from deficits in the bioengineering of the device itself because identical AMC-BAL bioreactors not exposed to toxin-laden plasma were able to support autologous hepatocytes with high levels of function over the same period.
Having observed this effect in the context of one type of artificial organ researchers interested in other devices with a cellular component would probably be well advised to determine whether metabolites that can induce senescence are present at elevated levels in the tissue fluids of patient groups they seek to treat.For example, Niwa and Shimizu (2012) demonstrated that indoxyl sulphate (derived from dietary tryptophan converted to indole in the gut) which circulates in plasma in vivo is taken up by human tubular kidney epithelial cells in vitro where it elevates ROS generation triggering a p53 dependent senescent state.In light of our findings the potential for processes such as this to impact the development of bioartificial kidneys seems obvious.
Our demonstration that both the growth arrest and altered biosynthetic aspects of SILT can be prevented with either resveratrol, dihydroresveratrol (a major in vivo metabolite) or several novel analogs has potential practical utility.The doses which we have used are both achievable and well tolerated by humans in vivo and resveratrol is currently undergoing phase 3 clinical trials for conditions as diverse as COVID-19 infection and mild cognitive impairment.Some BALS have also reached this stage but their performance has been disappointing to date.For example, a particularly promising BAL, the extracorporeal liver assist device (ELAD) utilised a HepG2 subclone in its bioreactors but in a phase 3 clinical trial for severe alcohol hepatitis (Thompson et al., 2018) failed to improve overall survival relative to the control group (51% vs 49.5%).However, if SILT was primarily or even partially responsible for this failure then it would be relatively simple to undertake more focused phase 2 or 3 studies in which a pre-existing BAL design using HepG2 cells was paired with resveratrol-mediated protection from senescence.The likelihood of a positive outcome could be further enhanced by the recruitment of patient subgroups less likely to produce high levels of SILT by pre-screening for the levels of the various hepatotoxic metabolites circulating in their plasma.
The mechanisms by which SILT occurs are clearly of significant interest but are likely to be quite varied within liver failure patients depending on the precise concentrations of the various hepatotoxic compounds.For example, toxic hydrophobic bile acids, such as deoxycholic acid conjugates, generate reactive oxygen species and induce hepatocyte apoptosis and thus can almost certainly trigger senescence (Granato et al., 2003).This pro-apoptotic effect can be blocked by nicotinic acid or nicotinamide (Crowley et al., 2000) which activate SIRT1 and blunted in the presence of high levels of bilirubin which acts as a radical scavenger (Granato et al., 2003).However, the toxicity and pro or anti-oxidant capacity of bilirubin varies widely depending on the pH, partial pressure of oxygen and the presence of other compounds (Qaisiya et al., 2017, Stocker et al., 1987, Sugita et al., 1986, Sugita et al., 1987).Taken together these findings might initially seem to favor a broad model in which oxidative stress is the principle initiator of SILT and the SIRT1 axis the primary mediator of protection (Cantó and Auwerx, 2012).
However, the use of our unique panel of resveralogues provides evidence firstly, that partial or total protection can be obtained regardless of whether SIRT1 is activated (since our V29 compound is capable of protecting but does not activate SIRT1) and secondly that simple antioxidant capacity is not the primary means by which protection occurs.
Although not a primary rationale for their selection, the compound series we have used includes resveralogues which lack the phenolic functionality that is essential to antioxidant and radical scavenging activity (Birar et al., 2020).If direct radical scavenging were the major mechanism by which protection occurs then the extent of protection would scale with the scavenging capacity of the resveralogue used as the protecting compounds.This does not occur but the availability of closely structurally related compounds allows us, in principle, to identify the major pathways triggering SILT (which could include indirect protection from oxidative stress) by a combination of next generation sequencing and exposure to clinically reflective mixtures of metabolic hepatotoxins.The potential now exists both for the development of BALS systems that are far more resistant to SILT and for the selection of patient subgroups considerably more likely to benefit from their use.Thus, despite many past failures, there is every possibility that functional bioartificial liver systems may become a clinical reality in the near future.

Fig. 1 .
Fig. 1. a. Evaluation of HepG2 Cell Viability Post Etoposide Treatment.The viability of HepG2 cells was assessed following exposure to 10 and 20 µM etoposide for 24 and 48 h of treatment.The percentage of viable cells was determined immediately after the removal of treatment using an MTT assay.Results are presented as the mean ± SD (n=3) and were analysed by Two-way ANOVA Bonferroni's multiple comparisons tests (***p<0.001compared to the control).Fig. 1b.Effect of 10 µM and 20 µM etoposide on the growth fraction of HepG2 cells after incubation for 48 h.The percentage of EdU-positive nuclei expressed as the Mean ± SD of three independent experiments using One-Way ANOVA with Bonferroni's Multiple Comparison Test (***p< 0.001).Fig. 1h compared to untreated control cultures.The data are expressed as the Mean ± SD of three independent experiments (***p< 0.001).Fig. 1d.Quantification of p53 positive cells.P53 staining was performed in duplicate and at least 100 cells were inspected to determine the percentage of p53-positive cells.The data are expressed as the Mean ± SD of three independent experiments using the Test (****p< 0.0001) (ET: etoposide) Fig. 1e.Effect of etoposide treatment on p53 gene expression in HepG2 cells.The expression level of p53 in etoposide treated and non-treated HepG2 cells was analysed by RT-PCR.Transcripts were normalised to GAPDH and are shown as fold change over control levels.Data was analysed using an unpaired t-test (*p<0.05).Fig. 1f.Senescence-associated (SA) β-galactosidase (β-gal) staining in HepG2 cell after treatment with etoposide (ET).(**p<0.01)Fig. 1g.Secretion of albumin by HepG2 cultures grown on 3-dimensional (3-D) HEMA-MBA-Alginate (HMA) and standard 6-well tissue culture plates (2-D) compared with identical populations treated with 10 µM etoposide for 48 h (ET).Data are the mean ±SD from three independent cultures.(*p˂0.05)n=3 Fig. 1h.Urea production by HepG2 cells cultured on 3-dimensional (3-D) HEMA-MBA-Alginate (HMA) and standard 6-well tissue culture plates (2-D) compared with identical populations treated with 10 µM etoposide for 48 h (ET).Each sample was treated with 1 mM ammonium chloride for 24 h to stimulate urea production.Data represent the mean ± SD from three independent cultures using One-way ANOVA Bonferroni's multiple comparison test.(***p˂0.001).

Fig. 2 .
Fig. 2. a.The number of viable cells on each HEMA-MBA-alginate cryogel was measured by MTT assay one day after removing treatment.Data were compared using One-Way ANOVA with Bonferroni's multiple comparison test.n=3 B. Live and dead staining of cryogels seeded with HepG2 cells after 6 h treatment with liver toxins and cytokines treatment (LT: liver toxins).The scale bar is 750 µm.Fig. 2b.Effect of a liver toxin cocktail (LT) either alone or supplemented with proinflammatory cytokines (LT + cytokines) on the growth fraction of HepG2 cells as measured by EdU incorporation after 6 h of exposure to the cocktail.Controls are either untreated HepG2 populations (Control) or identical cultures treated with etoposide as previously described (ET).The percentage of EdU-positive nuclei is expressed as the Mean ± SD of three independent experiments using One-Way ANOVA with Bonferroni's Multiple Comparison Test (****p< 0.0001).Fig. 2c.Effect of a liver toxin cocktail (LT) either alone or supplemented with proinflammatory cytokines (LT + cytokines) on the growth fraction of HepG2 cells after 6 h of exposure as measured by detection of pKi67.The percentage of Ki67 positive nuclei and data represent the average and standard deviation of three independent experiments using one-way ANOVA Bonferroni's multiple comparison test (***P value<0.001)Fig. 2d.Effect of liver toxins (LT) on gene expression of senescence-associated markers in hepG2 cells was measured by qPCR.Data were normalised to GAPDH and expressed as fold change over control levels.Data was analysed using an unpaired t-test (*p<0.05).Fig. 2e.Effects on the growth fraction of HepG2 cells as measured by EdU incorporation of 6 h of exposure to either a mixture of hepatoxic metabolites alone (LT), liver toxins in combination with proinflammatory cytokines (LT+cytokines) or single proinflammatory cytokines alone.The results are expressed as the Mean ± SD of three independent experiments using One-Way ANOVA with Bonferroni's Multiple Comparison Test (*p˂0.05,****p< 0.0001).

Fig. 4
Fig. 4. a.Effect of a liver toxin cocktail (LT) either alone or supplemented with 2 µM or 5 µM resveratrol (RSV) on the growth fraction of HepG2 cells in two dimensional culture as measured by EdU incorporation after 6 h of exposure to the hepatotoxic mixture.The controls are untreated HepG2 populations.The data are mean ± SD of EdU positive cells using One-Way ANOVA with Bonferroni's Multiple Comparison Test (***p< 0.001, **p<0.01,n=3).Fig. 4b.Effect of a liver toxin cocktail (LT) either alone or supplemented with 2 µM or 5 µM resveratrol (RSV) on the growth fraction of HepG2 cells in three-dimensional (HEMA-MBA-alginate cryogel) culture as measured by EdU incorporation after 6 h of exposure to the hepatotoxic mixture.The controls are untreated HepG2 populations.The data are mean ± SD of EdU positive cells using One-Way ANOVA with Bonferroni's Multiple Comparison Test (***p< 0.001,**p<0.01,n=3).Fig. 4c.Effect of a liver toxin cocktail (LT) either alone or supplemented with 2 µM or 5 µM resveratrol (RSV) on the growth fraction of HepG2 cells in two-dimensional culture as measured by Ki67 immunoreactivity after 6 h of exposure to the hepatotoxic mixture.The controls are untreated HepG2 populations.The data are mean ± SD of EdU positive cells using One-Way ANOVA with Bonferroni's Multiple Comparison Test (***p< 0.001, n=3).Fig. 4d.Effect of a liver toxin cocktail (LT) either alone or supplemented with 2 µM or 5 µM resveratrol (RSV) on the growth fraction of HepG2 cells in 3-dimensional (HEMA-MBA-alginate cryogel) culture as measured by Ki67 immunoreactivity after 6 h of exposure to the hepatotoxic mixture.The controls (cc) are untreated HepG2 populations.The bars show mean ± SD of EdU positive cells using One-Way ANOVA with Bonferroni's Multiple Comparison Test (**p<0.01,n=3).Fig. 4e.Quantification of SA-β-gal staining following a 6-hour exposure to liver toxins.Cells were assessed for the percentage of SA-β-gal positive staining three days post-treatment.SA-β-gal staining was performed in triplicate and at least 100 cells were inspected to determine the percentage of SA-β-gal positive cells.Statistical analysis performed using One-way ANOVA (***p< 0.001, ****p<0.0001).Results are mean with SD from N = three biological replicates (LT: Liver toxins, RSV: resveratrol).

Fig. 5 .
Fig. 5. a.Effect of a liver toxin cocktail on the growth fraction of HepG2 cells in two-dimensional culture as measured by EdU incorporation after 6 h of exposure to the hepatotoxic mixture either alone or liver toxins (LT) or supplemented with 5 µM concentrations of resveratrol (RSV), a major in vivo metabolite dihydroresveratrol (DIHRSV) or three novel resveralogues one of which (V29) has no capacity to activate SIRT1.The controls are untreated HepG2 populations.The data are mean± SD of EdU positive cells using one-way ANOVA Bonferroni's multiple comparison test ***Pvalue<0.001**P value<0.01(n=3).Fig. 5b.SA-β-gal Staining of HepG2 Cells in two-dimensional culture seven Days following exposure to hepatotoxins either alone (LT), or supplemented with 5 µM concentrations of V34 and V29.The untreated HepG2 populations serve as controls.The data represent the mean ± SD of SA-β-gal positive cells, analysed using a one-way ANOVA Bonferroni's multiple comparison test, with statistical significance denoted as ***P value<0.001(n=3).Fig. 5c.Effect of a liver toxin cocktail (LT) on the production of albumin by HepG2 cells cultured on three-dimensional matrices (HEMA-MBA-Alginate) either after 6 h of exposure to the hepatotoxic mixture either alone or liver toxins (LT) or supplemented with 5 µM concentrations of resveratrol (RSV), a major in vivo metabolite dihydro-resveratrol (DIHRSV) or the resveralogue V29 which has no capacity to activate SIRT1.Bars show the mean± SD of albumin production compared to liver toxin treatment using one-way ANOVA Bonferroni's multiple comparison test **p˂0.01,n=3) Fig. 5d.Effect of a liver toxin cocktail (LT) on the production of urea by HepG2 cells cultured on three-dimensional matrices (HEMA-MBA-Alginate) either alone, or liver toxins (LT) or supplemented with 5 µM concentrations of resveratrol (RSV), dihydroresveratrol (DIHRSV) or V29.Bars show the mean± SD of urea production compared to liver toxin treated HepG2 populations using one-way ANOVA Bonferroni's multiple comparison test (***p˂0.001,**p˂0.01,*p<0.05)(n=3).

Table 2
Primers for real-time RT-PCR.