A perfusion incubator liver chip for 3D cell culture with application on chronic hepatotoxicity testing

Abstract

Liver chips have been developed to recapitulate in vivo physiological conditions to enhance hepatocyte functions for assessing acute responses to drugs. To develop liver chips that can assess repeated dosing chronic hepatotoxicity, we need to ensure that hepatocyte functions be maintained at constant values over two weeks in stable culture conditions of sterility, temperature, pH, fluidic-flow of culture media and drugs. We have designed a perfusion-incubator-liver-chip (PIC) for 3D cell culture, that assures a tangential flow of the media over the spheroids culture. Rat hepatocyte spheroids constrained between a cover glass and a porous-ultrathin Parylene C membrane experienced optimal mass transfer and limited shear stress from the flowing culture media; maintained cell viability over 24 days. Hepatocyte functions were significantly improved and maintained at constant values (urea, albumin synthesis, and CYP450 enzyme activities) for 14 days. The chip act as an incubator, having 5% CO2 pressure-driven culture-media flow, on-chip heater and active debubbler. It operates in a biosafety cabinet, thus minimizing risk of contamination. The chronic drug response to repeated dosing of Diclofenac and Acetaminophen evaluated in PIC were more sensitive than the static culture control.

Full text

1
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
www.nature.com/scientificreports
A perfusion incubator liver chip for
3D cell culture with application on
chronic hepatotoxicity testing
Fang Yu1,2, Rensheng Deng1, Wen Hao Tong1,2, Li Huan1, Ng Chan Way2, Anik IslamBadhan1,
Ciprian Iliescu1,7,8,9 & Hanry Yu
1,2,3,4,5,6
Liver chips have been developed to recapitulate in vivo physiological conditions to enhance hepatocyte
functions for assessing acute responses to drugs. To develop liver chips that can assess repeated dosing
chronic hepatotoxicity, we need to ensure that hepatocyte functions be maintained at constant values
over two weeks in stable culture conditions of sterility, temperature, pH, uidic-ow of culture media
and drugs. We have designed a perfusion-incubator-liver-chip (PIC) for 3D cell culture, that assures a
tangential ow of the media over the spheroids culture. Rat hepatocyte spheroids constrained between
a cover glass and a porous-ultrathin Parylene C membrane experienced optimal mass transfer and
limited shear stress from the owing culture media; maintained cell viability over 24 days. Hepatocyte
functions were signicantly improved and maintained at constant values (urea, albumin synthesis,
and CYP450 enzyme activities) for 14 days. The chip act as an incubator, having 5% CO2 pressure-
driven culture-media ow, on-chip heater and active debubbler. It operates in a biosafety cabinet, thus
minimizing risk of contamination. The chronic drug response to repeated dosing of Diclofenac and
Acetaminophen evaluated in PIC were more sensitive than the static culture control.
‘Liver-on-a-chip’ models have attracted much attention since they recapitulate some in vivo tissue structures
and functions, biochemical cues and mechanical microenvironment. ese models permit the study of drug
interactions in vitro and could be alternatives to animal experimentation1,2. Liver chip models utilize minimal
cells and reagents, allowing more tests to be performed using human cells of limited quantity. e most common
‘Liver-on-a-chip’ models for drug testing are for testing acute drug toxicity3–5. ey have not been adapted to
long-term cell culture and chronic toxicity testing, due to the more pronounced problems of contamination, clog-
ging and bubble accumulation in the chips that deteriorate cell functions over extended culture period of weeks6,7.
Contamination of bacteria, mycoplasma or fungi is a common problem in long-term cell culture. Two stages
are critical in ensuring sterility in long-term cell culture: 1) initial chip assembly and cell seeding, 2) exposure to
contaminants during the extended cell culture over weeks. ere are a few strategies to prevent contamination in
the stage 1 of chip assembly, such as high temperature treatment or autoclave, Gamma irradiation or UV treat-
ment, and ltration of uid8,9. Proper sterile techniques in primary cell isolation from animals can ensure sterility
of the seeded cells in the chips10. Stage 2 is more problematic if we need to repeatedly disconnect and reconnect
the media/drug reservoir to the cell-containing chips. For monitoring cell responses under a microscope, we
also frequently move the chips in and out of the incubator. To minimize such frequent movements and connect/
disconnect in the non-sterile incubators, we integrate all the perfusion chip accessories into a single perfusion
incubator chip (PIC) such that the entire cell culture process is performed in the sterile biosafety cabinet.
1Institute of Bioengineering and Nanotechnology, A*STAR, The Nanos, 04-01, 31 Biopolis Way, Singapore, 138669,
Singapore. 2NUS Graduate School for Integrative Sciences and Engineering, Centre for Life Sciences (CeLS), 28
Medical Drive, Singapore, 117456, Singapore. 3MechanoBiology Institute, National University of Singapore, T-Lab,
5 A Engineering Drive 1, Singapore, 117411, Singapore. 4Department of Physiology, National University of Singapore,
MD9 03-03, 2 Medical Drive, Singapore, 117597, Singapore. 5Singapore-MIT Alliance for Research and Technology, 1
CREATE Way, #10-01 CREATE Tower, Singapore, 138602, Singapore. 6MechanoBiology Institute, National University
of Singapore, T-Lab, 5 A Engineering Drive 1, Singapore, 117411, Singapore. 7National Institute for Research and
Development in Microtechnologies, IMT-Bucharest, Bucharest, 077190, Romania. 8Academy of Romanian Scientists,
Splaiul Independentei nr. 54, sector 5, Bucharest, 050094, Romania. 9Bigheart, National University of Singapore,
MD6, 14 Medical Drive, #14-01, Singapore, 117599, Singapore. Correspondence and requests for materials should be
addressed to C.I. (email: ciprian.iliescu@imt.ro) or H.Y. (email: hanry_yu@nuhs.edu.sg)
Received: 1 June 2017
Accepted: 7 September 2017
Published: xx xx xxxx
OPEN
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
2
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
In long-term microuidic culture systems, dissociated gas from the culture media oen results in bubble
accumulation in the microuidic channels. is leads to clogging and dead volumes. Bubble traps commonly
used in microuidic systems can prevent large bubbles from entering the culture chamber; however, they cannot
prevent the formation of dissociated gas or tiny bubble accumulation inside the culture chambers. us, we have
integrated an active debubbler in the PIC to remove bubbles from the chamber in real time.
A proper design of a chronic liver toxicity-testing chip should have the following features:
• Organotypic 3D cellular architecture11–15;
• Good mass transfer, allowing diusion of O2 and nutrients from media to the spheroids and removing the
metabolites and by-products from the proximity of the spheroids1,16–18;
• Maintenance of mechanical forces and limited shear stress to the cells1,7,18;
• Stable cell culture conditions: temperature, pH, sterility and optimized culture media composition1,7,18;
• Ease of handling cells and replenishing media7;
Here we report a method and device for long-term 3D hepatocyte culture, which maintains the cell viability
over 3 weeks and functions over 2 weeks, and its application on chronic hepatotoxicity testing. An essential
aspect is that the design of the device assures a tangential ow over the cell culture, removing the metabolites and
by-products from the proximity of the cells and refreshing the cell environment. A constrain spheroids 3D rat
hepatocytes cell culture model was used for the testing. In this model, ultrathin Parylene C porous membrane
constrained the hepatocyte spheroids formed underneath; it protects the hepatocytes from the shear stress while
its relatively large pore size (20 µm) assures a good mass transfer. e membrane also immobilises the spheroids
and prevents them from colliding into each other, which frequently happens on non-adhesive surfaces19 and
ligand-modied lms20. PIC with integrated temperature, pH and bubble control provides contamination-free
and clogging-free environment to maintain rat hepatocytes viability and metabolic activities for 2–3 weeks. We
applied the PIC for acute and chronic repeat dosing drug safety testing using two model-drugs diclofenac and
acetaminophen (APAP). PIC addressed the various issues aecting the robustness and performance of liver chips
for chronic drug testing.
Results
PIC Structure and Functionality. Device structure and microuidic setup. e perfusion incubator liver
chip (PIC) was designed combining a rigid, well dened and reusable glass/silicon structure with the advantages
of PDMS assemblies (elasticity and gas permeable). e chip structure consists of three main elements illustrated
in Fig.1a:
• A glass/silicon structure containing a 3D microuidic circuit, the cell culture chamber, a bubble trap chamber
and a heater (Fig.1b). Nanoport microuidic connectors were mounted on the glass/silicon die. e micro-
uidic circuits were engraved on both sides of silicon die while the glass die assures the sealing of the bottom
microuidics circuit. On the other side of the glass die a heater is printed.
• A cell culture support (in our case a 100 µm-thick glass coverslip and a Parylene C membrane mounted on
a Silicon ring). e constrained spheroid structure can be inserted and removed from the culture chamber.
• A PDMS/glass sealing structure closes the microuidic circuit. e PDMS structure also acts as a debubbler
to absorb the air bubbles trapped when assembling the chip and generated during the culture. A cross-section
of the chip illustrating the functions of the PDMS/glass die is presented in Fig.1c.
e main materials of the PIC are silicon and glass. e main advantage is their biocompatibility21,22, while
having a hydrophilic surface of the microuidic elements, there is less absorption of proteins and drug molecules.
is is an important aspect for drug screening applications, especially when working with lower drug concen-
tration23. e silicon structure assures a uniform temperature of the media, due to its good thermal conductivity.
Moreover, the device can be reused to generate reproducible experimental results; and can be sterilized by owing
ethanol. For testing purposes, the silicon/glass structure was fabricated with dierent depths of the cell culture
chamber (0.5 mm, 1 mm, 2 mm and 3 mm). When the cell culture chamber is clamped with a so deformable
PDMS cover, the high rigidity of silicon enables water-tight sealing of the microwell.
To minimize the risk of microbial contamination, the PIC was kept inside a biosafety cabinet at all times. To
achieve this, we integrated a heater on chip. e heater was connected in series with a temperature controller and
a thermocouple. Using a second thermocouple, the temperature of the cell culture media in the cell culture cham-
ber and at the outlet was calibrated. e measured temperature is continuously adjusted to the set point of the
temperature controller. e integration of the heater on the chip avoids the use of the incubator. Pressurized CO2
was used to drive media perfusion through the chip (Supplementary1). CO2 dissolves in water forming carbonic
acid, which keeps pH of the media constant.
We integrated bubble trap and active debubbler on chip to remove bubbles accumulated in the culture cham-
ber. We avoided the use of external bubble traps to reduce the complexity of the system (more uidic connectors)
and surface area in contact with the cell culture media. Cells exposed to air bubbles are subjected to higher shear
stress due to the stretching force at liquid-air interface. Large bubbles can cause cell death. As shown in (Fig.1c),
the debubbler consists of a 70 µm-thick PDMS membrane (gas permeable) bonded to a PDMS molded chamber
with pillars that support the membrane. e PDMS structure is connected to external vacuum (through a pres-
sure controller). Gas bubbles trapped in the microwell can diuse through the PDMS membrane due to negative
pressure in the vacuum chamber while culture media remains inside the culture chamber. Time-lapse images
show that the vacuum eliminated the bubbles inside the microwell in 30 min (Supplementary2).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
3
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
Fabrication of microuidic glass/silicon structure. e glass/silicon structure was fabricated in four versions with
varying depth of the cell culture chamber (0.5 mm, 1 mm, 2 mm and 3 mm). We will describe the fabrication
process for the device with the depth of the chamber of 1mm. A 4” silicon wafer, 1mm-thick, with the crystallo-
graphic orientation <100> was used for the fabrication of the microuidic circuits. Main steps of the fabrication
process are illustrated in Fig.2. A similar process is described in23.
e silicon wafer was cleaned in piranha (H2SO4/H2O2 in ratio 2/1) in a quartz tank at 120 °C for 20 min,
rinsed in DI water and spun dried. A 2 µm-thick thermal SiO2 layer (wet process) was grown on the Si surface in
a furnace (Tystar, USA) at 1050 °C for 11 hours (Fig.2a). e pattern with the microuidic circuit was transferred
on the SiO2 layer (bottom side of the wafer) using a 2 µm-thick photoresist mask (AZ7217-Clariant) and a clas-
sical RIE etching process in CHF3/O2 using and (SPTS reactor) – Fig.2b. Aer the RIE process, the photoresist
mask was removed in an ultrasonic tank with NMP at 70 °C for 30 min, rinsed in DI water and spun dry. e
opposite microuidic circuit was aligned and imprinted in a similar manner on the other side of the Si wafer
(Fig.2c). A 10 µm-thick photoresist mask, having the pattern of the etch-through features (inlet outlet holes,
via-holes, cell culture chamber and bubble-trap chamber) was aligned and applied on the bottom of the wafer.
rough this mask, trenches were etched in Silicon (400 µm in depth) using a classical Bosch process (on an
Alcatel 101SE ICP Deep RIE) (Fig.2d). e photoresist mask was striped using O2 plasma process in the same
equipment, the SiO2 mask was then revealed. Another anisotropic etching of silicon was performed through the
SiO2 mask (100 µm deep). It generated the microuidic circuit on the bottom of the Si wafer (Fig.2e). A Teon
layer was uniformly deposited in the Deep RIE system using C4F8 chemistry for protection of the microuidic
structure during deep RIE etching from the opposite side of the wafer. In the next step (Fig.2f), the process
described in Fig.2d (400 µm deep trenches using Bosch process through a photoresist mask) was repeated, this
time from the topside of the wafer. Aer removing the photoresist mask in O2 plasma and temporary bonding
onto a dummy Si wafer with wax24, the microuidic circuit from the top side of the wafer was dened using the
same Bosch process till etch-through holes were completed (Fig.2g). Aer wax removing in NMP ultrasonic tank
and rinsing with DI water, the remaining SiO2 masks (top and bottom side of the wafer) were removed in BOE
(Buer Oxide Etch) solution, rinsed in DI water and cleaned in piranha. A 200nm-thick dry SiO2 layer was grown
on the surface of Si wafer in a furnace. is SiO2 layer assures a hydrophobic surface of the microuidic structure.
e silicon wafer is then anodically bonded on a 4”, 500 µm-thick Pyrex glass wafer (Corning 7740) (Fig.2h). e
heater of the chip consists of a Cr/Au metal deposition (40 nm/1 µm)(CHA e-beam evaporator). It was dened
using a classical photolithographic process (AZ7217) and Cr/Au wet etching solutions (Fig.2i). A passivation
layer SiO2/SiC (PECVD) was deposited on the heater and patterned using a photolithographic process and RIE
process (Fig.2j). is layer assures electrical isolation and chemical protection of the heater. Finally, the wafer
was diced. Nanoport microuidic connectors (Upchurch) (Fig.2k) and an electrical connector were mounted on
the chip (Fig.2l).
Figure 1. Schematic of the PIC chip. (a) 3D view with the PIC. A glass/silicon structure containing a 3D
microuidic circuit, the cell culture chamber, a bubble trap chamber as well as a heater (b) bottom view of
the chip’s layout illustrating the microuidic circuit, the cell culture chamber, the bubble trap and the heater,
(c) cross-section of the PIC illustrating the structure of the bubble trap. It consists of a 70 µm-thick PDMS
membrane (gas permeable) bonded to a PDMS molded chamber with pillars that support the membrane. e
PDMS structure is connected to external vacuum (through a pressure controller). e gas bubbles trapped in
the microwell can diuse through the PDMS membrane due to negative pressure in the vacuum chamber while
culture media remains inside the culture chamber. (d) Top and bottom view of the PIC.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
4
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
PIC System optimization. Flow simulation. Figure3 shows the proles of ow velocity and O2 concen-
tration at dierent depths in the cell culture chamber. Although the velocity is high near the inlet and outlet, it
becomes relatively low at the membrane plane due to the large cross-sectional area of the cell culture chamber.
is may help to lower the shear stress on the cells under the porous membrane. e O2 concentration decreases
along the ow direction due to the O2 consumption by the cells at the bottom.
e decrease along the depth direction is less signicant. is may be attributed to the shallowness of the cell
culture chamber (H = 0.5 mm) in which the O2 transfer from the top to bottom is relatively fast. Figure4a and b
show the averaged O2 concentration and shear stress at the membrane plane. e simulation was performed for dif-
ferent depth of the bioreactor (0.5, 1, 2 and 3 mm). With the increasing ow rate, more O2 is carried into the system
by the culture medium while the consumption rate is not signicantly aected. As a result, the O2 concentration at
the membrane level also increases, although the change is less signicant when the ow rate is beyond 0.1 mL/h.
On the other hand, for an increased depth of the cell culture chamber a lower O2 concentration (at the same ow
rate) was achieved, which is not surprising due to the corresponding higher resistance for O2 transfer. Moreover, the
shear stress on the membrane can increase almost linearly with the increasing ow rate, as shown in Fig.4b. It can
be observed from the slope of the lines that a shallow depth corresponds to a high shear stress. For example, at the
same ow rate, the shear stress for H = 0.5 mm is about 38 times higher than that for H = 3 mm. A typical enigma in
designing a cell culture chamber is to achieve a compromise between the requirement for mass transfer and the pro-
tection of cells from shear stress. A shallow chamber operating at high ow rates may provide better nutrient supply
and waste removal than a deeper one at low ow rates. However, the former usually generates higher shear stress
on the cells than the latter, as shown by the simulation results. In an environment with either insucient nutrients
and O2 or too much shear stress, the cells cannot exhibit proper metabolism or even become unviable. With the help
from the porous membrane, used in this study to shield cells from most of the shear stress, it is possible to design
a shallow chamber (e.g. H = 0.5 mm) and operate it at a relatively high ow rate (e.g. Q = 0.1 mL/h), in an eort to
provide sucient O2 to the cells for their metabolic activities. We will further demonstrate that the cell viability can
be aected by the shear stress at high ow rates (Fig.5a).
e O2 equilibrium is aected by not only the supply and transfer but also the consumption rate. Figure4c shows
the O2 level at the membrane plane with dierent number of cells cultured in the chamber. Obviously, a larger num-
ber of cells correspond to higher O2 consumption, resulting in a lower oxygen level in the region with cells.
Figure 2. Main steps of the PIC fabrication process: (a–l) fabrication of the Si/glass structure (legend T = top of
the wafer, B = Bottom of the wafer) (m–r) fabrication of the PDMS/glass cover.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
5
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
Optimization of the owing conditions. Flow rate was optimized with ow simulation and veried by selecting a
ow rate at which mass transfer is maximized without too much shear stress damage to the cells to ensure the best
performance of the on-chip culture. Hepatocyte spheroids were cultured under dierent ow rates: 0.02 mL/h,
0.06 mL/h, 0.1 mL/h, 0.2 mL/h and 0.4 mL/hr. MTS assay and urea assay is performed aer 3 days of experiment
to evaluate hepatocyte viability and function for dierent ow rates. To compare with static cultures, MTS activ-
ity of spheroids in PIC was measured in 48 well-plate, aer transferring the coverslips from the PIC back to the
well-plate. e results are presented in Fig.5a (hepatocytes viability) and Fig.5b (urea production). e spheroid
viability in the chip and collagen sandwich were benchmarked against the cell viability of freshly isolated hepat-
ocytes seeded in the 48 well-plate. Due to cell loss in the process of spheroid formation and transfer of coverslip,
the viability presented are around 30–40% of control. e working range is between 0.06 and 0.2 mL/h, having an
optimum ow rate at 0.1 mL/h.
Cell culture chamber depth aects cell viability and function. e inuence of chamber depth was studied.
Theoretically, a tangential flow over the surface of the membrane helps the removal of the metabolites and
by-products and refresh the media around the cells. Moreover, the concentration of the O2 and nutrients decrease
with the depth of the cell culture chamber (Fig.6a), so we expect an optimal hepatocyte function and viability can
be identied at optimal depth. We experimented with dierent depth of the cell culture chamber: 0.5 mm, 1 mm,
2 mm and 3 mm in order to appreciate the relevance of this parameter. e experiments are presented in detail in
Supplementary3 the cells cultured in chamber with 0.5 mm depth showed the highest secretion level.
Experimental validation of oxygen concentration simulation results. At the ow rate of 0.1 mL/h, the calculated
O2 consumed by the cells is 7.0%, 7.8%, 9.2% and 10.5% (from O2 level in the fresh media) for the depth of
0.5 mm, 1.0 mm, 2.0 mm and 3.0 mm, respectively (Fig.4a). is trend similar to the experimental observations in
Supplementary3: cells cultured in thinner cell culture chambers had better cell viability and metabolic functions
due to better supply of O2 and nutrients. We tested the drop in O2 concentration in culture media aer cell cul-
turing using uorescence signal of platinum-octaethyl-porphyrin (PtOEP) dye (method described previously25).
e O2 concentration can be quantied because of PtOEP’s oxygen-induced phosphorescence quenching eect.
e results are shown in Supplementary4 indicating a drop in O2 concentration of ~10%.
Experimental validation: temperature of the cell culture media in PIC. e temperature of the cell culture media
was veried in the owing conditions (0.02 to 0.4 mL/h) using a thermocouple. e thermocouple was placed
Figure 3. Simulation of the ow velocity and O2 concentration at dierent depths of the cell culture chamber.
e shear stress is uniform in the cell culture chamber due to the large cross-sectional area of the bioreactor.
is may help to lower the shear stress on the cells under the porous membrane. e O2 concentration decreases
along the ow direction due to the O2 consumption by the cells.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
6
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
(using a modied PDMS cover) inside the cell culture well as well as in the outlet microuidic connector. ere
was no dierence between the indication of the temperature controller and the measurements, showing good
thermal transfer of the design system.
Absorption of drugs to the PIC. e accuracy of drug testing experiments performed in microuidic devices
depends on achieving a consistent and reproducible concentration of drug molecules in the solution. However, in
microuidic devices with PDMS components, hydrophobic molecules tend to be absorbed to the walls of micro-
uidic devices, which reduces their concentration and aect the accuracy of drug testing results. We quantied
the absorption of drugs to the tubing and microuidic chip, the results are presented in Supplementary5. e
surface area of the PDMS cover (~450 mm2) exposed to the media in the PIC was much smaller than the surface
area in the tubing (~2300 mm2). Majority of the drug absorption was found to be attributed to the tubing. Initially,
aer 30 minutes of perfusion, 74% of the APAP can be retrieved from the outlet, only 52% of the diclofenac can
be retrieved. Aer 48 hours of perfusion, drug molecules absorbed to the surface starts to saturate, 98% of APAP
and 95% of diclofenac can be retrieved. Diclofenac absorbs more easily to PDMS surface than APAP because of its
hydrophobicity and higher log P value. To avoid cross-contamination of the drugs due to absorption, the PDMS
cover of the chip and the tubing were not reused in the experiment.
Comparison with a laminar ow bioreactor. To benchmark the performance of the PIC against other perfusion
culture systems, we compared the functions of cells in a bioreactor reported previously26 and the PIC (in both
cases we use the same constrained spheroids cell culture model). e functions are compared with the cells cul-
tured in 3 other control congurations. Monolayer setting refers to seeding hepatocyte directly to cell culture
plates. Spheroids on AHG has the same substrate modication method and cell-culture conguration as PIC but
cultured in multi-well plates. In this setting, hepatocytes are seeded on a low-adhesion substrate (glass covers-
lips modied with poly(ethylene glycol) (PEG) and galactose ligand 1-O-(60-aminohexyl)-D-galactopyranoside
(AHG, MW 279)) to promote spontaneous spheroid formation20. We previously demonstrated that hepatocyte
Figure 4. CFD simulations (a) oxygen concentration (b) average of shear stress (c) normalized oxygen
concentration for dierent depth of the cell culture chamber.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
7
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
spheroids cultured on AHG-modified substrate exhibited improved polarity and liver-specific functions27.
Collagen sandwich conguration is the industry standards that has oen been recommended in FDA guidelines
for in vitro hepatotoxicity testing28.
We measured the urea synthesis and albumin secretion levels in both perfusion culture systems over 8 days
(Fig.6). e cells cultured in PIC and the bioreactor produced similar levels of urea while both higher than static
culture as expected. e albumin production in PIC is signicantly higher than the laminar ow bioreactor. ese
results indicate that tangential ow that characterize the PIC structure is essential in maintaining cell functions.
We used experiment success rate to gauge the robustness of PIC compared with bioreactor. All the 12 exper-
iments conducted with PIC and collagen sandwich were successful without contamination. 9 out of 12 experi-
ments with spheroids cultured on AHG were successful, with 3 experiments having extensive cell detachments
aer 7 days of culture. Only 3 out of 12 experiments were successful for laminar ow bioreactor, mostly because
of bacterial contamination issue during the culture. All the experiments were conducted by the same operator.
Maintenance of cell function in PIC. Long-term cell viability on chip. e PIC system for hepatocyte
constrained spheroid culture has been tested for long-term cell culture for 24 days. Optical images taken during
this period of the constrained spheroids are presented in Fig.7a. At day 24 a live/dead staining was performed.
e images presented in Fig.7b demonstrates the viability of the cultured cells can be maintained for 24 days.
Maintenance of hepatocyte dierentiated function. To investigate whether the PIC can support chronic hepato-
toxicity testing of drugs, we studied whether the hepatocyte functions can be maintained over 14 days. e results
of urea secretion and albumin synthesis are presented in Fig.8a and b respectively.
Urea synthesis in spheroids on AHG increased compared to collagen sandwich culture (Fig.8a); the increase
was more signicant than the increase in albumin secretion (Fig.8b). e urea produced by hepatocytes cultured
in collagen sandwich conguration is 87.7, 79.4, 38.4, 34.5, 35.5, 28.9 µg/106 cells/day on day 2, 4, 6, 8, 12 and 14
respectively. By comparison, the urea secreted by spheroids on AHG ranged from 282.3, 199.9, 161.6, 121.6, 153.6,
178.3 µg/106 cells/day on day 2, 4, 6, 8, 12 and 14 respectively. Urea secretion further improved in PIC: 393.7,
181.5, 241.5, 243.76, 132.5, 203.4 µg/106 cells/day on day 2, 4, 6, 8, 12 and 14 respectively. Notably, hepatocytes
cultured in PIC secreted about 7 times the amount of urea compared with collagen sandwich.
Hepatocytes cultured in PIC showed signicant increase in albumin secretion compared to collagen sandwich
throughout 14 days of culture (Fig.8b). e amount of albumin secreted by collagen sandwich cultured hepat-
ocytes on day 2, 4, 6, 8, 12 and 14 were 19.5, 39.6, 43.4, 38.4, 20, 7.3 µg/106 cells/day respectively. e amount of
albumin secreted by spheroids on AHG in the same period were 21.8, 29.9, 26.4, 15.3, 8.2 and 3.3 µg/106 cells/day
Figure 5. Optimization of the PIC system: (a) hepatocyte viability at dierent ow rates (b) urea production
during 3 days culture for dierent ow rates. e working ow rate range is between 0.6 and 2 mL/h, the
optimum point is 0.1 mL/h.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
8
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
on day 2, 4, 6, 8, 12 and 14 respectively. Albumin secretion capacity of spheroids on AHG improved in perfusion
culture. e amount of albumin secreted by hepatocytes in PIC on day 2, 4, 6, 8, 10, 12 and 14 were 35.9, 50.7,
48.2, 45, 37.17, 18.3 µg/106 cells/day respectively.
We quantied the gene expression levels of CYP1A2, CYP2B1/2 and CYP3A2 in PIC and compared with
monolayer and collagen sandwich culture. ese CYPs are rat homologs of human CYP1A2, CYP2B6, and
CYP3A4 respectively. ey have important roles in drug metabolism. CYP1A2 is the second most abundant
CYPs in human body and is inducible by many of the carcinogens and aromatic hydrocarbons. CYP2B6, although
only account for 5% of total CYP component, is responsible for the metabolism of more than 25% of drugs in
the market. CYP3A4 is the most abundant in human liver and responsible for metabolism of two-thirds of all
marketed drugs.
e activities of CYP1A2, CYP2B1/2 and CYP3A2 enzymes were measured using their respective specic
substrates: phenacetine, bupropion and midazolam (Fig.9a). e quantity of their respective metabolite acetami-
nophen, OH-bupropion and 1′-OH-midazolam were measured with LC-MS. We found that hepatocytes cultured
in spheroid model shows enhanced CYP1A2 activity; the amount of acetaminophen (metabolite of phenace-
tine metabolized by CYP1A2) produced by collagen sandwich cultured hepatocytes was measured at 49.2ng/106
cells/90 min, while hepatocytes in PIC produced 195.1ng/106 cells/90 min on Day 8.
On day 14, the amount of acetaminophen measured in collagen sandwich and PIC were 26.9ng/106
cells/90 min and 156.4ng/106 cells/90 min respectively.
e CYP2B1/2 activity of hepatocytes was quantied by measuring the amount of OH-bupropion metabolized
from bupropion (Fig.9b). On day 8, the amount of OH-bupropion produced by collagen sandwich was 17.78 ng/
106 cells/90 min, lower than spheroids on AHG, which produced 32.49 ng/106 cells/90 min of OH-bupropion.
By comparison, Hepatocytes in PIC produced signicantly higher OH-bupropion: 67.32 ng/106 cells/90 min on
day 8. On day 14, the amount of OH-bupropion produced by collagen sandwich and spheroids on AHG were
12.38 ng/106cells/90 min and 22.54 ng/106cells/90 min respectively. e amount of OH-bupropion produced by
PIC cultured hepatocytes was slightly higher at 76.56 ng/106cells/90 min but this is not statistically signicant.
To quantify the activity of CYP3A2, we measured the amount of 1′OH-midazolam metabolized by CYP3A2
from midazolam (Fig.9c). On day 8, the amount of 1′-OH-midazolam produced by collagen sandwich and sphe-
roids on AHG were 57.2ng/106 cells/90 min and 140.6 ng/106 cells/90 min respectively. e amount of 1′- OH
midazolam produced by spheroids in PIC was higher at 167.2ng/106 cells/90 min. On day 14, collagen sandwich
Figure 6. Comparison of cellular functions over 8 days in PIC and a laminar ow bioreactor. (a) e cells
cultured in PIC and the bioreactor produced higher levels of urea compared with static cultures. (b) Albumin
production in PIC is signicantly higher than other groups. Results were obtained from triplicate measurements
(n = 3), *p ≤ 0.05. Data from representative experiments are presented, whereas similar trends were seen in
multiple trials.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
9
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
and spheroids on AHG produced 30.6ng/106 cell/90 min and 74.6ng/106 cell/90 min of 1′-OH-midazolam. By
comparison, spheroids in PIC produced 164.5 ng/106 cells/90 min of 1′-OH-midazolam.
Application of PIC-cultured hepatocytes in drug safety testing. We studied the hepatocyte spheroids under static
and perfusion (PIC) conditions for two weeks for the evaluation of chronic drug eects. Acute toxicity and
chronic toxicity of two drugs: Diclofenac and Acetaminophen (APAP) were tested and compared. Diclofenac
causes rare but signicant cases of serious hepatotoxicity including liver necrosis, jaundice, fulminant hepatitis
with and without jaundice, and liver failure29. Diclofenac toxicity is dicult to be evaluated due to the prob-
lematic estimation of the dose response. Its hepatotoxic eects are not easily reproducible in current animal
models, indicating idiosyncratic toxicity of diclofenac29. erefore, the liver function should be monitored on
long-term therapy with diclofenac since increased AST/ALT levels are observed in clinical treatment30. In vivo,
delayed diclofenac induced hepatotoxicity usually occurs29. Previous studies showed that in vitro, only high con-
centrations of diclofenac (>200 µM) induce acute toxicity30,31 in primary cultures of human hepatocytes. us
diclofenac repeatedly dosed over 2 weeks in our in vitro PIC was tested. For APAP as a most frequently used pain
killer in the world, there is large body of literature characterizing its safety in vitro and in vivo32–34. Both acute and
chronic overdosing of APAP can cause liver toxicity and liver failure32.
Acute toxicity (24 h) of diclofenac is presented in Fig.10a. In this experiment, we observed a signicant toxic
eect at 500 µM and 1 mM concentration. At the highest concentration (1 mM), the cell viability signicantly
decreased to 10%. ere is no relevant dierence between PIC and collagen sandwich culture for diclofenac
acute toxicity. e IC50 was calculated to be 495.61 µM and 493.66 µM respectively. For APAP acute toxicity, we
observed that cells cultured in PIC were more sensitive to APAP treatment than the cells cultured in collagen
sandwich (Fig.10c). e IC50 of APAP was 22.51 mM and 38.57 mM for PIC and collagen sandwich.
Figure 7. (a) Phase contrast images with constrained spheroids during 24 days cell culture. Cell viability can be
maintained up to day 24; (b) live/dead assay at day 24, showing viable cells with intact spheroid structure. Scale
bar represents 100 μm.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
10
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
Chronic drug toxicity of diclofenac and APAP were evaluated by measuring the viability of hepatocyte sphe-
roids perfused with dierent drug concentrations for 14 days. For chronic hepatotoxicity, static cultured hepato-
cytes (Fig.10b), the lowest cell viability was found at 100 µM diclofenac. In the perfusion culture, spheroids were
exposed to repeated dosing of diclofenac every 2 days. e perfusion system is more sensitive for testing chronic
drug response of diclofenac: the IC 50 is 50.86 µM for PIC and 121.53 µM for the collagen sandwich control. At
100 µM diclofenac chronic dosage, cell viability is reduced to less than 40% in PIC aer 14 days of culture, whereas
100 µM diclofenac dosing doesn’t show a signicant acute toxicity. Similarly, cells cultured in PIC is more sensitive
to chronic APAP toxicity (Fig.10d): at 1 mM concentration, cell viability was reduced to 21% in PIC, whereas
cells cultured in the collagen sandwich control had 32% viability compared to the DMSO control. e IC 50 is
2.391 mM for PIC and 3.059 mM for the collagen sandwich control.
Discussion
In vitro liver cell models based on conventional cell culture platforms such as 2D monolayer and collagen sand-
wich can support hepatocyte function for a week. 3D cell culture congurations can enhance functions to higher
level with their own limitations such as mass transfer and heterogeneous environments etc. Bell et al. proposed
a 3D primary human hepatocyte model for in vitro study of liver functions and diseases and drug response35
Perfusion culture models have been developed to regulate the soluble microenvironment and can sustain hepato-
cyte function in a conned space for a longer period, such as suspension spheroids perfusion-stirred bioreactor36,
encapsulated spheroids bioreactor37 and laminar ow spheroids bioreactor27. Various microuidic chips have
been developed to provide physiological shear stress and liquid to cell ratio within the microuidic compart-
ments for a few days38–41. Carraro et al.42 proposed a perfusion hepatocyte cell culture having a “vascular-like”
microuidic structure in hepatocyte culture seeded on a nanoporous membrane. Vernetti et al.43 proposed a
3D human liver model organized in four-cell sandwich structure (hepatocytes, endothelial cells, stellate cells
and Kuper cells). Lee et al.44 fabricated a PDMS device which mimics the liver anatomy. e device features
an articial liver sinusoid with an articial barrier layer mimicking endothelial barrier layer. We have also inte-
grated cell lines and primary cells into 3D microuidic channels to evaluate the cellular response to drug expo-
sure4,26,27,45–47. However, these devices have not been utilized in chronic drug safety testing applications due to the
Figure 8. Synthetic function of hepatocytes cultured in monolayer, collagen sandwich, constrained spheroid
in static culture and constrained spheroid on PIC over 14 days. (a) Hepatocytes cultured in PIC produced
signicantly higher amount of urea compared with monolayer and collagen sandwich. (b) Albumin secretion
is signicantly higher in PIC than other groups. Results were obtained from triplicate measurements (n = 3),
*p ≤ 0.05. Data from representative experiments are presented, whereas similar trends were seen in multiple
trials.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
11
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
deterioration of hepatocyte functions and contamination problems associated with repeating dosing of drugs to
the cells. A number of chips have been developed for long-term culture48–52. Wagner et al.53 combined traditional
PDMS microuidic device with transwell inserts and created a multi-organ chip capable of supporting long-term
cell culture. Maschmeyer and colleagues54 developed a PDMS multi-chamber chip for long-term co-culture of
4 tissue types. Recently, Kang et al.49 presented a long-term transwell microuidic device using PDMS micro-
channels and polyethyleneterephthalate (PET) membranes to mimic the liver sinusoid structure. ese devices
have the advantage of adjusting uid ow and controllable liquid to sample ratio, enabling prolonged cell cul-
ture. However, these PDMS devices all rely on traditional incubator and involve complex uidic connections to
operate. e drawback of these devices is that when they undergo frequent transfer of devices from incubator
to microscope for observation, changing media and adding drug doses, the risk of contamination and bubble
generation- once the chip is reinserted in the microuidic setup- becomes higher. Our PIC is such a device that
integrates incubator functions such as temperature and pH control, and with active debubbler for 3D long-term
culture and systematically ensure contamination-free long-term cell culture for repeatedly dosed drug testing.
is PIC system doesn’t need to be inserted in an incubator system, it works independently in a biosafety cabinet.
Figure 9. Higher CYPs enzymatic activity for PIC cultured hepatocytes compared with monolayer, collagen
sandwich and Spheroids on AHG. (a) CYP1A2 activity, measured by the amount of acetaminophen metabolized
from phenacetine; (b) CYP2B1/2 activity, measured by the amount of OH-bupropion metabolized from
bupropion; (c) CYP3A2 activity, measured by the amount of 1′-OH-midazolam metabolized from midazolam.
Results were obtained from triplicate measurements (n = 3), *p ≤ 0.05. Data from representative experiments
are presented, whereas similar trends were seen in multiple trials.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
12
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
It enabled long-term maintenance of 3D hepatocyte spheroid culture. We tested cell viability for 24 days, while
the cell function was veried for 2 weeks. e integration of all the elements such as temperature controller and
active debubbler allowed constrained spheroid model to be cultured in a stable microenvironment for weeks.
Heaters have been used in cell culture platforms adapted to robotic systems and real-time imaging systems55–58.
We applied this concept to our long-term cell-culture chip and implemented a thermocouple and a temperature
controller to achieve precise maintenance of temperature during prolonged culture period. We have created a
closed microenvironment for cell culture and manipulation inside a sterile biosafety cabinet, eliminating the
need of an incubator. Moreover, the tangential ow allowed refreshing of media, removing of by-products and
protection of the spheroids cell culture from shear stress.
Microuidic cell-culture systems typically need to be pre-conditioned and sterilized before cell seeding59.
Bubbles can form at uidic connections or generated from the dissociated gas in the media. A great deal of
care is needed during the operation of microuidic chips to ensure bubble-free condition in the cell culture. In
the long-term cell-culture chip, the chance of bubble accumulation in the microuidic channels is even higher,
amplifying the eects of bubbles to perfusion ow and cell function. Bubble traps60–62 are usually used to prevent
bubbles from entering cell culture area of the microuidic chips. When the bubble trap is lled with bubbles
over weeks, additional bubbles can still enter cell culture areas. erefore, we integrated a bubble trap and an
active-debubbler to trap and remove the bubbles generated during long-term culture.
A long-term cell-culture system must also maintain a microenvironment with favourable mass transport
and limited shear stress. In this study, we used an established hepatocyte culture model: the constrained sphe-
roid model, due to its superior ability to maintain cell function and minimize cell loss27. Here, we further opti-
mized the spheroid seeding and perfusion culture process. We pre-formed the spheroids on glass coverslips and
constrained them with parylene C ultra-thin porous membrane in 48-well plates. Aer spheroid stabilization,
they are transferred to the PIC chip to take advantage of the perfusion ow. We demonstrated improved and
well-maintained activity of hepatocytes’ CYP1A2, CYP2B1/2 and CYP3A2 enzymes. We validated the use of PIC
for acute and chronic drug toxicity studies. We found that for APAP, the acute toxicity response is more sensitive
in the PIC compared to static culture. Similar APAP toxicity testing results63 were reported in perfusion biore-
actor: perfusion of APAP resulted in a shi in dose response such that 20 mM of APAP in perfusion culture lead
to signicant acute toxicity compared with 40 mM in static culture. e reason is that, the metabolic activity is
elevated in perfusion culture microenvironment. Hepatocytes can metabolize APAP faster and produce higher
quantity of its reactive intermediate NAPQI, which causes free-radical damage of cellular structures64. We further
validated the use of our PIC chip for chronic drug testing by evaluating hepatotoxicity aer 2 weeks’ exposure to
diclofenac and APAP and obtained a more sensitive chronictoxicity response in the perfusion-cultured spheroid
model compared with the static cultured collagen sandwich control. Chronic toxicity of diclofenac is more sensi-
tive in the PIC compared with static culture. Diclofenac chronic toxicity is largely attributed to one of its phase II
metabolite: diclofenac-acylglucuronide due to covalent binding65. Phase II metabolism requires longer time for
metabolite production than primary metabolites. erefore at 100 mM concentration, diclofenac doesn’t show
acute toxicity but results in chronic toxicity.
e PIC is currently a low throughput chip that addressed the contamination issues associated with long-term
culture of cells in 3D to preserve high levels of cell function for drug testing applications. In the near future, such
Figure 10. Comparison of acute and chronic dose response to (a,b) Diclofenac and (c,d) APAP between
collagen sandwich and PIC (day 1 and day 14), Quantied MTS levels were normalized to respective control
conditions. Results were obtained from triplicate measurements (n = 3). Data from representative experiments
are presented, whereas similar trends were seen in multiple trials.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
13
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
in vitro model could be used to study the complex spatiotemporal cues that govern long-term changes in cell fates.
e PIC can be further developed into high-throughput format, to involve multiple cell types in co-culture, and
to support industry-scale drug screening applications.
In conclusion, we presented a method and a system for robust long-term hepatocytes constrained spheroid cul-
ture for chronic drug testing. e PIC integrates a heater and temperature controller to control temperature, CO2
pressure-driven culture-media ow to control pH, and active debubbler on chip, in a biosafety cabinet, to minimize
the risk of contamination. e tangential ow over the parylene membrane (that stabilize the spheroids and limit the
shear stress) assures a controlled microenvironment. e PIC maintained the cell functions for 2–3 weeks, support-
ing repeated dosing chronic drug testing. is PIC is a robust long-term 3D perfusion culture platform that can be
used also for culturing dierent cell types for other sub-acute and chronic drug testing applications.
Methods
Flow simulation. As a powerful tool based on the transport theories of momentum, mass and energy,
computational uid dynamics (CFD) simulation yields the proles of shear force and O2 concentration in this
study. e geometry and mesh of the ow domain were constructed using GAMBIT 2.3.16, and the created
mesh was introduced into FLUENT 6.2.16 for calculation. e ow was considered as laminar, steady and
three-dimensional, with the operating temperature of 37 °C. e inlet and outlet of the bioreactor were mod-
elled as “velocity-inlet” and “outow”, respectively, while the walls were considered as no-slip. In addition to
the Navier–Stokes equations, a species transport equation was solved to calculate the oxygen concentration
in the bioreactor. e concentration of dissolved O2 at the inlet was set as 6.08e-3 kg/m3 according to its sol-
ubility in plasma (α) at 37 °C. e zone occupied by the hepatocytes under the membrane was modelled as a
“source term” for O2, from which the dissolved oxygen was consumed at the Oxygen Uptake Rate (OUR) of
the hepatocytes66.
α
α
=× +
OURV
c
cK
/
/
(1)
m
m
where c is the local O2 concentration; Vm is the maximum rat hepatocytes O2 consumption rate; and Km is the
half-saturation constant for hepatocytes. e parameters were taken from Table 1 of the reference66. Since c is an
unknown variable in Eq. (1), a user-dened-function (UDF) was used to determine the O2 source in an iteration
manner. A simple calculation shows that OUR decreases by only 3% when c decreases by 50% from its solubility,
indicating that the O2 consummation rate is not very sensitive to the change of local O2 concentration in this
range. e number of hepatocytes, unless otherwise specied, was set at 23,200, identical to that used in experi-
ments. e calculation process was continued till all the residues for continuity, momentum and species fell below
10−7.
Rat hepatocytes isolation and spheroid pre-culture. We used a constrained spheroid model
established previously27 as the hepatocyte cell culture model. Hepatocytes aggregated in spheroids are able
to retain polarity and form functional bile canaliculi67. Functional parameters such as albumin and urea
production, CYP450 activities are also improved compared with collagen monolayer19,68 and collagen sand-
wich culture69,70.
To obtain hepatocyte spheroids, we isolated primary rat hepatocytes from male Wistar rats of 250–300 g in
weight using a two-step collagenase perfusion method37. Animals were handled according to the Institutional
Animal Care and Use Committee (IACUC) protocol. Isolation protocol was reviewed and approved by the
Biological Resource Center (BRC) Institutional Animal Care and Use Committee. Freshly isolated hepatocytes
were seeded onto collagen-coated 48-well plate or PEG-galactose modied glass coverslips at 1 × 105 cells/cm2 to
pre-form the spheroids27. e glass coverslips were modied according to the protocol27. e cells were cultured
in William’s E medium supplemented with 1 mg/mL BSA, 10 ng/mL EGF, 0.5 μg/mL insulin, 5 nM dexametha-
sone, 50 ng/mL linoleic acid, 100 units/mL penicillin and 100 μg/mL streptomycin; and incubated with 37 °C, 5%
CO2, 95% humidity. Cell seeding glass coverslips 10mm in diameter were purchased from Paul Marienfeld GmbH
& Co.KG (Lauda-Königshofen, Germany). Silane-PEG-COOH, MW 5000 was purchased from Nanocs Inc.
(New York, USA). 1-O-(6′-aminohexyl)-D-galactopyranoside (AHG, M.W. 279), the galactose ligand was syn-
thesized in house as reported previously and veried by NMR spectrum20,71. Wafers used for chip fabrication
were purchased from Bonda Technology (Singapore). Other chemicals were purchased from Sigma-Aldrich
(Singapore) unless otherwise stated. Parylene C membranes used for sandwich constrained spheroid culture
were fabricated and surface modied27. Aer 1 day in static culture, the coverslips with attached spheroids were
manually moved to PIC using sharp point forceps and covered with parylene C membrane. e PIC was sealed
by clamping the PDMS-glass cover on top of the culture chamber. To evaluate the robustness of the PIC, we
conducted cell culture experiments in cell culture plates, in a laminar ow bioreactor previously developed26
and in the PIC in parallel. Six independent 14-day cell culture experiments were performed in duplicates for
each setup.
Collagen sandwich culture. We used both the hepatocyte monolayer and the collagen sandwich as stand-
ard controls for hepatotoxicity testing of drugs, in order to compare with the results achieved from the PIC. e
bottom collagen-coating substrate was prepared by adding 40 µL neutralized collagen type I solution (Invitrogen,
Palo Alto, USA) onto the 10mm glass coverslip before incubation at 37 °C overnight for gelation. Hepatocytes
were seeded on the collagen-coated coverslip at 1 × 105 cells/cm2 density; they were incubated for 1 h for full
attachment before media replenishment and then cultured for 24 h. e culture medium was removed; subse-
quently, 40 µL of un-gelled collagen was overlaid on top of the cells. Gelation of the collagen overlay was allowed
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
14
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
to occur at 37 °C for 3 h before fresh medium was replenished. e collagen sandwich culture was placed in the
incubator with daily change of culture media.
Urea and albumin synthesis measurement. 1 mL of culture media were collected for urea and albumin
synthesis measurements each day. Urea content in the culture media was measured using Urea Nitrogen Kit
(Stanbio Laboratory, USA). Albumin concentration was measured using Rat Albumin ELISA Quantication Kit
(Bethyl Laboratories Inc., USA). e absolute urea and albumin amounts were calculated based on media volume
collected and was normalized against the cell number by the end of culture.
Diclofenac and APAP concentration measurement. To determine the amount of diclofenac and
APAP absorbed to the PIC and tubing, concentration of diclofenac72 and APAP73 were measured by colorimetric
method using a Tecan innite m200 plate reader (Tecan, Switzerland).
Cell viability staining. e viability of hepatocytes was visualized by a dual staining method. 2 uorescent
nuclear dyes: Propidium iodide (PI) (Molecular Probes, USA) and Calcein AM (Molecular Probes, USA) were
used to stain the necrotic and viable cell population. e staining was done by moving the glass coverslip from the
PIC to a microscope slide. e cells were stained by incubating in 100 μl of 25 mg/mL PI (30 minutes) and 1 μM
Calcein AM (30 minutes); the samples were washed 3 times with 1X PBS before imaging. All stained samples were
imaged with a confocal microscope at 488 nm and 543 nm excitation.
Liquid chromatography-mass spectrometry (LC-MS) measurement. For CYPs specific activ-
ity analysis, hepatocytes were cultured for 8 or 14days before adding CYP-specific substrates diluted in
Krebs-Henseleit buffer (KHB) and incubated for 1.5 hours. Hepatocytes in perfusion cultured were removed
from the bioreactor and transferred to multi-well plates for this experiment. The samples were collected and
stored at −80 °C until LC-MS measurement. To conduct the LC-MS measurement, 50 µL of 100ng/mL inter-
nal standards were added to the samples and the mixture was dried using Techne® Sample Concentrator
(Techne, UK). The dried residues were reconstituted using 100 µL of methanol containing 0.1% formic
acid. The supernatant was then analysed using LC-MS system (LC: 1100 series, Agilent, US; MS: LCQ Deca
XP Max, Thermo Finnigan, US) with 100 × 3.0 mm onyx-monolithic C18 column (Phenomenax, USA) as
reported previously74.
Acute toxicity. e hepatocyte spheroids were tested for acute toxicity. 1 day aer culturing in PIC, e
cells were exposed to 4 concentrations of diclofenac (10–1000 μM) and APAP (10–40 mM) for 24 hours under
perfusion condition. DMSO vehicle controls were established by using medium supplemented with 1% DMSO
without drugs. Viability was tested aer removing the glass slides containing cells from PIC to a 48-well-plate. e
viability was assessed with MTS assay (Promega, USA).
Chronic toxicity. For the assessment of chronic toxicity of diclofenac and APAP, the hepatocytes were
exposed to four concentrations of diclofenac (1–100 μM) and APAP (0.1–10 mM). e selection of these con-
centrations range corresponds to the therapeutic serum concentration (Cmax) of diclofenac (6.4 μM)75 and APAP
(1 mM)76. Vehicle control (0.1% DMSO) was used. Serum free William’s E medium was used during drug treat-
ment. Repeated doses were given every 48 h for static culture upon medium change. e viability was assessed
with MTS assay (Promega, USA) upon 14 days aer seeding of the hepatocytes.
Statistical analysis. At least 3 experimental replicates were used for each data point. In each experiment, 2
chips were used to generate duplicated data. Statistical comparisons were performed using 2 way ANOVA. Results
are expressed as means ± standard error of the means (sem) of 3 independent experiments.
References
1. Huh, D., Hamilton, G. A. & Ingber, D. E. From 3D cell culture to organs-on-chips. Trends Cell Biol 21, 745–754, https://doi.
org/10.1016/j.tcb.2011.09.005 (2011).
2. Andersen, M. E. & rewsi, D. Toxicity testing in the 21st century: bringing the vision to life. Toxicol Sci 107, 324–330, https://doi.
org/10.1093/toxsci/fn255 (2009).
3. Anene-Nzelu, C., Wang, Y., Yu, H. & Liang, L. H. Liver tissue model for drug toxicity screening. Journal of Mechanics in Medicine
and Biolog y 11, 369–390 (2011).
4. Toh, Y.-C. et al. A microuidic 3D hepatocyte chip for drug toxicity testing. Lab Chip 9, 2026–2035 (2009).
5. Materne, E. M., Tonevitsy, A. G. & Marx, U. Chip-based liver equivalents for toxicity testing–organotypicalness versus cost-ecient
high throughput. Lab Chip 13, 3481–3495, https://doi.org/10.1039/c3lc50240f (2013).
6. Zheng, W., Wang, Z., Zhang, W. & Jiang, X. A simple PDMS-based microuidic channel design that removes bubbles for long-term
on-chip culture of mammalian cells. Lab on a Chip 10, 2906–2910 (2010).
7. Wu, M. H., Huang, S. B. & Lee, G. B. Microfluidic cell culture systems for drug research. Lab Chip 10, 939–956, https://doi.
org/10.1039/b921695b (2010).
8. Xu, W., Liang, L., Song, Z. & Zhu, M. Continuous ethanol production from sugarcane molasses using a newly designed combined
bioreactor system by immobilized Saccharomyces cerevisiae. Biotechnol Appl Biochem 61, 289–296 (2014).
9. Farid, S. S. Cost-eectiveness and robustness evaluation for biomanufacturing. BioProcess International 11, 20–27 (2013).
10. eese, J. A. & Byard, J. L. Isolation and culture of adult hepatocytes from liver biopsies. In Vitro 17, 935–940 (1981).
11. Zhang, C., Zhao, Z., Abdul ahim, N. A., van Noort, D. & Yu, H. Towards a human-on-chip: Culturing multiple cell types on a chip
with compartmentalized microenvironments. Lab on a Chip 9, 3185–3192 (2009).
12. Imura, Y., Sato, . & Yoshimura, E. Micro total bioassay system for ingested substances: assessment of intestinal absorption, hepatic
metabolism, and bioactivity. Anal Chem 82, 9983–9988 (2010).
13. S onntag, F. et al. Design and prototyping of a chip-based multi-micro-organoid culture system for substance testing, predictive to
human (substance) exposure. J Biotechnol 148, 70–75 (2010).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
15
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
14. Wu, M.-H., Huang, S.-B., Cui, Z., Cui, Z. & Lee, G.-B. A high throughput perfusion-based microbioreactor platform integrated with
pneumatic micropumps for three-dimensional cell culture. Biomed Microdevices 10, 309–319 (2008).
15. Yu, F. et al. On chip two-photon metabolic imaging for drug toxicity testing. Biomicrouidics 11, 034108 (2017).
16. Curcio, E. et al. Mass transfer and metabolic reactions in hepatocyte spheroids cultured in rotating wall gas-permeable membrane
system. Biomaterials 28, 5487–5497, https://doi.org/10.1016/j.biomaterials.2007.08.033 (2007).
17. Bhushan, A. et al. Towards a three-dimensional microuidic liver platform for predicting drug ecacy and toxicity in humans. Stem cell
research & therapy 4Suppl 1, S16, https://doi.org/10.1186/scrt377 (2013).
18. Webster, A., Greenman, J. & Haswell, S. J. Development of microuidic devices for biomedical and clinical application. Journal of
Chemical Technology & Biotechnology 86, 10–17, https://doi.org/10.1002/jctb.2482 (2011).
19. oide, N. et al. Formation of multicellular spheroids composed of adult rat hepatocytes in dishes with positively charged surfaces
and under other nonadherent environments. Exp Cell es 186, 227–235 (1990).
20. Ying, L. et al. Immobilization of galactose ligands on acrylic acid gra-copolymerized poly(ethylene terephthalate) lm and its
application to hepatocyte culture. Biomacromolecules 4, 157–165, https://doi.org/10.1021/bm025676w (2003).
21. otzar, G. e t al. Evaluation of MEMS materials of construction for implantable medical devices. Biomaterials 23, 2737–2750 (2002).
22. Ni, M. et al. Cell Culture on MEMS Platforms: A eview. Int J Mol Sci 10, 5411–5441 (2009).
23. Choudhury, D. et al. Fish and Chips: a microuidic perfusion platform for monitoring zebrash development. Lab on a Chip 12,
892–900 (2012).
24. Ilies cu, C., Taylor, H., Avram, M., Miao, J. & Franssila, S. A practical guide for the fabrication of microuidic devices using glass and
silicon. Biomicrouidics 6, 016505 (2012).
25. Wang, L. et al. Construction of oxygen and chemical concentration gradients in a single microuidic device for studying tumor cell-
drug interactions in a dynamic hypoxia microenvironment. Lab on a Chip 13, 695–705, https://doi.org/10.1039/c2lc40661f (2013).
26. Xia, L. et al. Laminar-ow immediate-overlay hepatocyte sandwich perfusion system for drug hepatotoxicity testing. Biomaterials
30, 5927–5936, https://doi.org/10.1016/j.biomaterials.2009.07.022 (2009).
27. Tong, W. H. et al. Constrained spheroids for prolonged hepatocyte culture. Biomaterials 80, 106–120, https://doi.org/10.1016/j.
biomaterials.2015.11.036 (2016).
28. Dunn, J. C., Tompins, . G. & Yarmush, M. L. Long-term in vitro function of adult hepatocytes in a collagen sandwich
conguration. Biotechnol Prog 7, 237–245 (1991).
29. Boelsterli, U. A. Diclofenac-induced liver injury: a paradigm of idiosyncratic drug toxicity. Toxicol Appl Pharmacol 192, 307–322
(2003).
30. Bort, ., Ponsoda, X., Jover, ., Gomez-Lechon, M. J. & Castell, J. V. Diclofenac toxicity to hepatocytes: a role for drug metabolism
in cell toxicity. J Pharmacol Exp er 288, 65–72 (1999).
31. Lauer, B., Tuschl, G., ling, M. & Mueller, S. O. Species-specic toxicity of diclofenac and troglitazone in primary human and rat
hepatocytes. Chemico-biological interactions 179, 17–24, https://doi.org/10.1016/j.cbi.2008.10.031 (2009).
32. L ane, J. E., Belson, M. G., Brown, D. . & Scheetz, A. Chronic acetaminophen toxicity: a case report and review of the literature. e
Journal of emergency medicine 23, 253–256 (2002).
33. Shear, N. H. et al. Acetaminophen-induced toxicity to human epidermoid cell line A431 and hepatoblastoma cell line Hep G2, in
vitro, is diminished by silymarin. Sin pharmacology: the ocial journal of the Sin Pharmacology Society 8, 279–291 (1995).
34. Jemnitz, ., Veres, Z., Monostory, ., obori, L. & Vereczey, L. Interspecies dierences in acetaminophen sensitivity of human, rat,
and mouse primary hepatocytes. Toxicol In Vitro 22, 961–967, https://doi.org/10.1016/j.tiv.2008.02.001 (2008).
35. Bell, C. C. et al. Characterization of primary human hepatocyte spheroids as a model system for drug-induced liver injury, liver
function and disease. Scientic reports 6, 25187 (2016).
36. Tostoes, . M. et al. Human liver cell spheroids in extended perfusion bioreactor culture for repeated-dose drug testing. Hepatology
55, 1227–1236, https://doi.org/10.1002/hep.24760 (2012).
37. Tostoes, . M. et al. Perfusion of 3D encapsulated hepatocytes–a synergistic eect enhancing long-term functionality in bioreactors.
Biotechnol Bioeng 108, 41–49, https://doi.org/10.1002/bit.22920 (2011).
38. Marx, U. et al. ‘Human-on-a-chip’developments: a translational cutting-edge alternative to systemic safety assessment and eciency
evaluation of substances in laboratory animals and man. Alternatives to Laboratory Animals-ATLA 40, 235 (2012).
39. Neužil, P., Giselbrecht, S., Länge, ., Huang, T. J. & Manz, A. evisiting lab-on-a-chip technology for drug discovery. Nature reviews
Drug discovery 11, 620–632 (2012).
40. Bhatia, S. N. & Ingber, D. E. Microuidic organs-on-chips. Nat Biotechnol 32, 760–772, https://doi.org/10.1038/nbt.2989 (2014).
41. Yu, F., Iliescu, F. S. & Iliescu, C. A Comprehensive eview on Perfusion Cell Culture Systems. Informacije MIDEM 46, 163–175
(2016).
42. Carraro, A. et al. In vitro analysis of a hepatic device with intrinsic microvascular-based channels. Biomedical microdevices 10,
795–805 (2008).
43. Vernetti, L. A. et al. A human liver microphysiology platform for investigating physiology, drug safety, and disease models.
Experimental biology and medicine 241, 101–114 (2016).
44. Lee, P. J., Hung, P. J. & Lee, L. P. An articial liver sinusoid with a microuidic endotheliallie barrier for primary hepatocyte
culture. Biotechnology and bioengineering 97, 1340–1346 (2007).
45. Zhan g , C. et al. e controlled presentation of TGF-β1 to hepatocytes in a 3D-microuidic cell culture system. Biomaterials 30,
3847–3853 (2009).
46. Toh, Y.-C. et al. A novel 3D mammalian cell perfusion-culture system in microuidic channels. Lab on a Chip 7, 302–309 (2007).
47. Zhu, L. et al. A vertical-ow bioreactor array compacts hepatocytes for enhanced polarity and functions. Lab on a Chip, doi:https://
doi.org/10.1039/C6LC00811A (2016).
48. Maschmeyer, I. et al. A four-organ-chip for interconnected long-term co-culture of human intestine, liver, sin and idney
equivalents. Lab on a Chip 15, 2688–2699, https://doi.org/10.1039/C5LC00392J (2015).
49. ang, Y. B. et al. Liver sinusoid on a chip: Long-term layered co-culture of primary rat hepatocytes and endothelial cells in
microuidic platforms. Biotechnol Bioeng 112, 2571–2582, https://doi.org/10.1002/bit.25659 (2015).
50. Materne, E.-M. et al. A multi-organ chip co-culture of neurospheres and liver equivalents for long-term substance testing. J
Biotechnol 205, 36–46 (2015).
51. Mei, J.-C. et al. ree-Dimensional Extracellular Matrix Scaolds by Microuidic Fabrication for Long-Term Spontaneously
Contracted Cardiomyocyte Culture. Tissue Engineer ing Part A 20, 2931–2941 (2014).
52. Le e, M. et al. Long-term, feeder-free maintenance of human embryonic stem cells by mussel-inspired adhesive heparin and collagen
type I. Acta Biomater 32, 138–148 (2016).
53. Wagner, I. et al. A dynamic multi-organ-chip for long-term cultivation and substance testing proven by 3D human liver and sin
tissue co-culture. Lab Chip 13, 3538–3547, https://doi.org/10.1039/c3lc50234a (2013).
54. Maschmeyer, I. et al. A four-organ-chip for interconnected long-term co-culture of human intestine, liver, sin and idney
equivalents. Lab Chip 15, 2688–2699, https://doi.org/10.1039/c5lc00392j (2015).
55. Hung, P. J. et al. A novel high aspect ratio microuidic design to provide a stable and uniform microenvironment for cell growth in
a high throughput mammalian cell culture array. Lab on a Chip 5, 44–48 (2005).
56. Futai, N., Gu, W., Song, J. W. & Taayama, S. Handheld recirculation system and customized media for microuidic cell culture. Lab
on a Chip 6, 149–154 (2006).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

www.nature.com/scientificreports/
16
ScIeNtIfIc RepoRTs | 7: 14528 | DOI:10.1038/s41598-017-13848-5
57. El-Ali, J., Sorger, P. . & Jensen, . F. Cells on chips. Nature 442, 403–411 (2006).
58. Stangegaard, M., Petronis, S., Jørgensen, A. M., Christensen, C. B. V. & Dufva, M. A biocompatible micro cell culture chamber
(μCCC) for the culturing and on-line monitoring of euaryote cells. Lab on a Chip 6, 1045–1051 (2006).
59. im, L., Toh, Y. C., Voldman, J. & Yu, H. A practical guide to microuidic perfusion culture of adherent mammalian cells. Lab Chip
7, 681–694, https://doi.org/10.1039/b704602b (2007).
60. Selley, A. M. & Voldman, J. An active bubble trap and debubbler for microuidic systems. Lab Chip 8, 1733–1737, https://doi.
org/10.1039/b807037g (2008).
61. Sung, J. H. & Shuler, M. L. Prevention of air bubble formation in a microuidic perfusion cell culture system using a microscale
bubble trap. Biomed Microdevices 11, 731–738 (2009).
62. Eddington, D. In-line microuidic bubble trap. Lab Chip-Chips & Tips, http://blogs. rsc. org/chipsandtips/2006/11/22/in-line-
microuidic-bubble-trap/, Accessed Nov-22 (2006).
63. A llen, J. W., hetani, S. . & Bhatia, S. N. In vi tro zonation and toxicity in a hepatocyte bioreactor. Toxicological sciences 84, 110–119
(2004).
64. McClain, C. J., Price, S., Barve, S., Devalarja, . & Shedlofsy, S. Acetaminophen hepatotoxicity: an update. Current gastroenterology
reports 1, 42–49 (1999).
65. retz-ommel, A. & Boelsterli, U. A. Mechanism of covalent adduct formation of diclofenac to rat hepatic microsomal proteins.
etention of the glucuronic acid moiety in the adduct. Drug metabolism and disposition 22, 956–961 (1994).
66. Zhang, S. et al. A robust high-throughput sandwich cell-based drug screening platform. Biomaterials 32, 1229–1241, https://doi.
org/10.1016/j.biomaterials.2010.09.064 (2011).
67. Abu-Absi, S. F., Friend, J. ., Hansen, L. . & Hu, W. S. Structural polarity and functional bile canaliculi in rat hepatocyte spheroids.
Exp Cell es 274, 56–67, https://doi.org/10.1006/excr.2001.5467 (2002).
68. Wu, F. J., Friend, J. ., emmel, . P., Cerra, F. B. & Hu, W. S. Enhanced cytochrome P450 IA1 activity of self-assembled rat
hepatocyte spheroids. Cell Transplant 8, 233–246 (1999).
69. Taahashi, ., Sonoda, H., Tabata, Y. & Hisada, A. Formation of hepatocyte spheroids with structural polarity and functional bile
canaliculi using nanopillar sheets. Tissue Eng Part A 16, 1983–1995, https://doi.org/10.1089/ten.TEA.2009.0662 (2010).
70. Saai, Y., Yamagami, S. & Naazawa, . Comparative analysis of gene expression in rat liver tissue and monolayer- and spheroid-
cultured hepatocytes. Cells, tissues, organs 191, 281–288, https://doi.org/10.1159/000272316 (2010).
71. Findeis, M. A. Stepwise synthesis of a GalNAc-containing cluster glycoside ligand of the asialoglycoprotein receptor. International
journal of peptide and protein research 43, 477–485 (1994).
72. Perez-uiz, T., Martinez-Lozano, C. & Sanz, A. & San Miguel, M. T. Flow extraction spectrophotometric method for the
determination of diclofenac sodium in pharmaceutical preparations. Journal of pharmaceutical and biomedical analysis 16, 249–254
(1997).
73. Shihana, F., Dissanayae, D., Dargan, P. & Dawson, A. A modied low-cost colorimetric method for paracetamol (acetaminophen)
measurement in plasma. Clinical toxicology 48, 42–46, https://doi.org/10.3109/15563650903443137 (2010).
74. Xia, L. et al. Tethered spheroids as an in vitro hepatocyte model for drug safety screening. Biomaterials 33, 2165–2176, https://doi.
org/10.1016/j.biomaterials.2011.12.006 (2012).
75. Hinz, B. et al. Bioavailability of diclofenac potassium at low doses. Br J Clin Pharmacol 59, 80–84, https://doi.org/10.1111/j.1365-
2125.2005.02226.x (2005).
76. Davern, T. J. et al. Measurement of serum acetaminophen-protein adducts in patients with acute liver failure. Gastroenterology 130,
687–694, https://doi.org/10.1053/j.gastro.2006.01.033 (2006).
Acknowledgements
is work was supported in part by Institute of Bioengineering and Nanotechnology, & grants from JCO, ASTAR;
MBI (through NRF and MOE), NMRC-CBRG, SMART-BioSyM, NUHS to H.Y. F.Y., N.C.W. and W.H.T. were
NGS scholars.
Author Contributions
F.Y. and C.I. conceived the experiments, F.Y., R.D., W.H.T., L.H., N.C.W., A.I.B. and C.I. conducted the
experiment(s), F.Y., W.H.T., L.H. and C.I. analysed the results. All authors reviewed the manuscript.
Additional Information
Supplementary information accompanies this paper at https://doi.org/10.1038/s41598-017-13848-5.
Competing Interests: e authors declare that they have no competing interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre-
ative Commons license, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons license and your intended use is not per-
mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
© e Author(s) 2017
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com