Low-dose acetaminophen induces early disruption of cell-cell tight junctions in human hepatic cells and mouse liver

Dysfunction of cell-cell tight junction (TJ) adhesions is a major feature in the pathogenesis of various diseases. Liver TJs preserve cellular polarity by delimiting functional bile-canalicular structures, forming the blood-biliary barrier. In acetaminophen-hepatotoxicity, the mechanism by which tissue cohesion and polarity are affected remains unclear. Here, we demonstrate that acetaminophen, even at low-dose, disrupts the integrity of TJ and cell-matrix adhesions, with indicators of cellular stress with liver injury in the human hepatic HepaRG cell line, and primary hepatocytes. In mouse liver, at human-equivalence (therapeutic) doses, dose-dependent loss of intercellular hepatic TJ-associated ZO-1 protein expression was evident with progressive clinical signs of liver injury. Temporal, dose-dependent and specific disruption of the TJ-associated ZO-1 and cytoskeletal-F-actin proteins, correlated with modulation of hepatic ultrastructure. Real-time impedance biosensing verified in vitro early, dose-dependent quantitative decreases in TJ and cell-substrate adhesions. Whereas treatment with NAPQI, the reactive metabolite of acetaminophen, or the PKCα-activator and TJ-disruptor phorbol-12-myristate-13-acetate, similarly reduced TJ integrity, which may implicate oxidative stress and the PKC pathway in TJ destabilization. These findings are relevant to the clinical presentation of acetaminophen-hepatotoxicity and may inform future mechanistic studies to identify specific molecular targets and pathways that may be altered in acetaminophen-induced hepatic depolarization.


Methods overview Electric Cell-substrate Impedance Sensing
We utilized the commercial system Electric Cell-substrate Impedance Sensing (ECIS-Zθ, Applied Biophysics, Troy, NY USA), first to characterize establishment of the organotypic (HepaRG co-culture 1 liver model on the microelectrodes, and then to study dose-/ timedependent effects (0-24 hours) of APAP on cellular parameters of TJ, cell-substrate adhesion, and membrane integrity. A technical workflow, outlining the HepaRG-based model, and subsequent impedance spectral data deconvolution, into biologically-relevant parameters of cell behaviour, are shown in Fig. 2 and Supplementary Figure 1. Parallel assessment of cellular morphology, phenotype, gene and protein expression, alone, or in the context of APAP hepatotoxicity in HepaRG cells, were used to provide correlation with impedancebased quantitative measurements, and liver biochip system validation. Furthermore, effects of both phorbol ester (PKC-activator which abrogates hepatic TJs integrity) and purified NAPQI, on HepaRG TJs were investigated. See Supporting Information (below) for methodological details of: HepaRG and PHHs cell culture on ECIS microelectrodes; hepatotoxicity assays; flow cytometry analysis; immunocytochemistry; morphological and ultrastructural assessment; and impedance spectral modeling data analysis.
Impedance-based cellular assays (IBCAs) represent an emerging technology in drug discovery, due to their high sensitivity and quantitative nature. Pioneered by Giaever and Keese 2 , impedance sensing can detect minute changes in the behaviour of cells directly cultured on micro-electrode arrays. Impedance sensing is utilized mostly as a global indicator of cellular status, called Cell Index, and has been applied across many fields of biology such as, cell-substrate attachment and spreading 3,4 , signal transduction 5 , cytotoxicity 6 , metastasis 7 , cardiology 8 , and regenerative medicine 9 . Very rarely however, is impedancesensing fully exploited in its spectroscopic form, to retrieve biologically meaningful electrical parameters, such as paracellular and transcellular resistance. Typically, IBCAs have found a niche application in blood-brain barrier permeability studies, where impedance sensing has provided real-time quantitative monitoring of the establishment or loss of cell-cell endothelial TJs 3 .

Culture of Primary Human Hepatocytes on ECIS microelectrodes and treatment with acetaminophen following rifampicin induction
ECIS 8W10E+ arrays were pre-coated with collagen, as per manufacturer's instructions (Biopredic Int.), before seeding with primary human hepatocytes (PHH). PHH were seeded at 200, 000 cells per well and incubated at 37 o C for 24h to allow for cell attachment. Plates were then connected to ECIS and placed in the incubator to quantitatively monitor PHH growth kinetics using impedance measurements. Cells were exposed to Rifampicin, a known CYP3A4 inducer, for 24h prior to a 24h APAP hepatotoxicity assay. The effect of 0, 5, 10 and 20mM APAP on PHH impedance measurements was then monitored for 24h. ECIS modelling was then performed to deconvolve the impedance into its different parameters, in order to ascertain the effect of APAP on tight junctions, cell adhesion and cell membrane integrity.

Hepatotoxicity assays: primary human hepatocytes (PHHs)
A parallel study was conducted using a collagen-coated 96 well plate (Corning™ BioCoat™ Collagen I) to correlate ATP (industry standard endpoint) and PrestoBlue (live-cell) assays, with the ECIS measurements. PHHs were seeded at 50,000 cells per well. Cells were induced using Rifampicin for 24 hours before APAP hepatotoxicity was tested at 0, 5, 10 and 20mM as described above. The plate was removed from the incubator to equilibrate to room temperature (20 o C). PrestoBlue solution was prepared in media using a 9:1 dilution and 100µl was added to each well in addition to 4 no-cell control wells. The plate was then incubated at room temperature for 30 minutes in the dark in order to preserve the fluorescent signal. After 30 minutes, the plate was read using GloMax: Protocol PrestoBlue. The plate was washed with 100µl media per well and 4 no-cell control wells were selected before an ATP assay was performed according to manufacturer's instructions (Promega. CellTitreGlo).

Supplementary Figure 1 Electric Cell-substrate Impedance Sensing Model
Schematic showing basic principles underlying ECIS measurements: Electric Cell-substrate Impedance Sensing (ECIS) is a method used to monitor and characterise cells cultured on top of microelectrodes. The impedance of the cells is measured by driving a small alternating current (AC) signal between the sensing (working) electrode, and a larger counter electrode. A lock-in amplifier is then used to measure the variations in the complex impedance, composed of real (resistive) and imaginary (capacitive) parts. As cells grow on top of the sensing electrode, they impede current flow resulting in an increase in the measured complex impedance. Hence, any change in cell morphology or behaviour is mirrored by the impedance measurements. With reference to Supplementary Fig. 1 outlining impedance spectral modeling (see Fig, 2 for comparison), the ECIS-Zθ system scans impedance measurements through various frequencies to recognize the current pathways, and translates spectral data (Z') by deconvolution, using the built-in mathematical model (ECIS-Zθ software), into biologically-relevant cell electrical parameters: Rb (cell-cell junctions), z-alpha (cellelectrode adhesion) and Cm (cell membrane capacitance). This is based on the fact that at low frequencies, the current flows underneath and in-between the cells, reflecting both relative tightness of cell-cell junctions, and the degree of cell adherence to the electrodes. At high frequencies, the current can capacitively couple through the cells, revealing data on intracellular properties and integrity of the cell plasma membrane.

Supplementary Figure 2 Hepatotoxicity assays following 24 hours APAP or PMA treatment
To assess hepatotoxic response of APAP or PMA in HepaRG cells and for comparison with impedance measurements (see Fig. 3, Fig. 4), dose-dependent hepatotoxicity was detected at 24 hours, by multiplexing ATP-depletion endpoint assay (a, c) with the PrestoBlue live-cell viability assay (b, d). Prestoblue and ATP values are expressed relative to the levels found in control cells, arbitrarily set at a value of 100%. ATP content decreased to only 50% of control values at 20 mM APAP (a); with minimal change observed at 5-10 mM. Cell metabolic activity, as measured by Prestoblue reduction (providing also a quantitative measure of viability and cytotoxicity), was more pronounced in sub-toxic (5 mM) and intermediate (10 mM) dose APAP, compared with ATP-depletion assay. PMA caused moderate increases of both ATP-content (c), and PrestoBlue live-cell viability (d), at 24 hours; possibly due to hormetic effects of the PKC activator, PMA.

Supplementary Figure 3 Real-time impedance monitoring (0-24 hours) of HepaRGbased liver-on-chip device following 24 hours APAP challenge: Non-normalized dose response
Effect of APAP on non-normalized impedance data in real-time. Comparative impedance spectral modeling data is plotted; expressed as absolute impedance values (termed, 'nonnormalized' data) for all parameters (Global, Z' (Ω); Rb (Ω.cm 2 ); z-alpha (Ω 0.5 cm); and Cm (µF/cm2). Absolute values, with statistics are shown in Supplementary Table 1. The doseand time-dependent decline in the overall resistance is shown in (a) while the effect of APAP on tight junctions, cell-substrate adhesion and cell membrane capacitance is illustrated in (b), (c) and (d), respectively. Both the non-normalized and normalized data (Fig. 3) showed a concentration-and time-dependent decrease in the resistance (Z') and the modeling parameters (Rb; z-alpha; Cm). Subsequently, data was normalized through dividing the impedance (or its deconvolved parameters) by its value at the APAP challenge starting point (t=0); see corresponding normalized data in NAPQI formation is the major cause of APAP hepatotoxicity, we therefore tested direct effects of different concentrations of NAPQI on TJs (Rb parameter) with impedance sensing. At higher dose NAPQI (500 µM), non-normalized resistance showed an approximate 50% reduction over 20 hours (a), indicating a decline in cellular health. An abrupt (0-2 hours) and sustained (0-20 hours) disruption of TJs occurred (b), with a 50% decrease in the impedance parameter (Rb) after 6 hours. Z-alpha also declined by 20% after 20 hours (c). Minimal effects recorded on both Rb and z-alpha (at 125 and 250 µM), may reflect detoxification of NAPQI by intrinsic stores of glutathione (GSH), as high doses of APAP are required to deplete mitochondrial GSH. The apparent abberrant effect on Cm at high dose NAPQI, could be due to the labile nature of NAPQI, formation (over hours) of protein adducts, causing lipid peroxidation of the cell membrane.

Supplementary Figure 6: Impedance measurements in HepaRG cells following 24 hour rifampicin
All APAP toxicity assays were preceded by an induction phase of rifampicin (CYP3A4 inducer). Rifampicin showed minimal effect on HepaRG cells as monitored by impedance biosensing. (a) Effect of rifampicin on resistance: Rifampicin induction did not have any deleterious effects on global cell health, as reflected by impedance measurements. (b) Similarly, Rifampicin induction did not cause any disruption to impedance-modeled parameters, Rb (cell-cell tight junctions), z-alpha (cell-substrate adhesion) or Cm (cell membrane integrity).

Supplementary Figure 7 Disruption of hepatic architecture in HepaRG cells following 24 hours APAP treatment: Co-localization F-actin-phalloidin/ E-cadherin immunofluorescence staining
Phase Minimal changes were observed in cell viability of primary human hepatocytes using either industry standard endpoint (ATP; panel a) or live-cell (Prestoblue; Panel b) hepatotoxicity assays, even at 20mM APAP.

Supplementary Figure 11 Hepatotoxicity assays for albumin and lactate production following 24 hours APAP treatment in HepaRG cells
To assess hepatotoxic response of APAP (0-40 mM) in HepaRG cells and for comparison with impedance measurements (see Fig. 3), dose-dependent hepatotoxicity was detected at 24 hours, by assessing albumin (A) and lactate (B) levels, as previously described 11,12 . Values for albumin shown are expressed relative to the levels found in control cells, arbitrarily set at a value of 100%. Values for lactate are in mmol/ L. In HepaRG cultures, albumin production decreased from 0.36 (untreated) to 0.04 (40 mM) ug/h/10 6 viable cells seeded. Conversely, lactate release increased sharply at 5mM APAP, and according to dose (5-15 mM), and plateaued at 20-40 mM.   (6, 12, 24 hours). At 20 mM APAP, the decline in resistance (Z') is highly significant even after only 6 hours (P<0.01), indicating reduction in global cellular health. After 12 hours, the decrease in Z' is significant for the other doses (P<0.01) and at the end of the 24 hours assay, the resistances reach their minimum values, close to that of the cell free electrode for high (10-20 mM) doses (∼300 Ω). The main effect of APAP is on the TJ parameter, Rb. At 6 h, highly significant decreases in Rb were measured for high dose (10-20 mM) APAP (P<0.01). At 12h, the decrease in Rb is significant even for the 5 mM dose with complete abolition of TJs at 10 and 20 mM APAP. Significant effects of APAP on cell-substrate adhesion (z-alpha), occurs later (>12 hours) than the effect on cell-cell TJs. At 6h, no significant decrease in adhesion is detected, while at 12h the 20 mM dose shows a significant decrease, and at 24 hours both the high doses (but not the low 5 mM dose), show a significant decrease in cell-substrate adhesion. The effect of APAP on the integrity of the cell membrane is only monitored for the high doses. It is more pronounced at 24 hours where Cm reaches zero, as membrane integrity is completely compromised, presumably with cell death and detachment. See corresponding Fig. 3.