Human hepatic in vitro models reveal distinct anti-NASH
potencies of PPAR agonists
Joost Boeckmans & Alessandra Natale & Matthias Rombaut & Karolien Buyl & Brent Cami &
Veerle De Boe & Anja Heymans & Vera Rogiers & Joery De Kock & Tamara Vanhaecke
Robim M Rodrigues
Received: 24 March 2020 /Accepted: 17 June 2020
# Springer Nature B.V. 2020
Abstract Non-alcoholic steatohepatitis (NASH) is a
highly prevalent, chronic liver disease characterized by
hepatic lipid accumulation, inflammation, and concom￾itant fibrosis. Up to date, no anti-NASH drugs have been
approved. In this study, we reproduced key NASH
characteristics in vitro by exposing primary human he￾patocytes (PHH), human skin stem cell-derived hepatic
cells (hSKP-HPC), HepaRG and HepG2 cell lines, as
well as LX-2 cells to multiple factors that play a role in
the onset of NASH. The obtained in vitro disease
models showed intracellular lipid accumulation,
secretion of inflammatory chemokines, induced ATP
content, apoptosis, and increased pro-fibrotic gene ex￾pression. These cell systems were then used to evaluate
the anti-NASH properties of eight peroxisome
proliferator-activated receptor (PPAR) agonists
(bezafibrate, elafibranor, fenofibrate, lanifibranor,
pemafibrate, pioglitazone, rosiglitazone, and
saroglitazar). PPAR agonists differently attenuated lipid
accumulation, inflammatory chemokine secretion, and
pro-fibrotic gene expression.
Based on the obtained readouts, a scoring system was
developed to grade the anti-NASH potencies. The
in vitro scoring system, based on a battery of the most
performant models, namely PHH, hSKP-HPC, and LX-
2 cultures, showed that elafibranor, followed by
saroglitazar and pioglitazone, induced the strongest
anti-NASH effects. These data corroborate available
clinical data and show the relevance of these in vitro
models for the preclinical investigation of anti-NASH
Keywords Non-alcoholic steatohepatitis (NASH) .
Peroxisome proliferator-activated receptor (PPAR) .
In vitro . Elafibranor. Pioglitazone . Saroglitazar
Non-alcoholic steatohepatitis (NASH) is a chronic
liver disease that is characterized by hepatic
steatosis, lobular inflammation, hepatocyte balloon￾ing, and fibrosis (Friedman et al. 2018). NASH
represents a prominent stage in the non-alcoholic
Cell Biol Toxicol
Tamara Vanhaecke and Robim M Rodrigues are equally
contributing senior authors.
Electronic supplementary material The online version of this
article contains
supplementary material, which is available to authorized users.
J. Boeckmans : A. Natale : M. Rombaut : K. Buyl :
B. Cami : A. Heymans : V. Rogiers : J. De Kock :
T. Vanhaecke :
. M. Rodrigues (*)
Department of In Vitro Toxicology and Dermato-Cosmetology,
Faculty of Medicine and Pharmacy, Vrije Universiteit Brussel,
Laarbeeklaan 103, 1090 Brussels, Belgium
e-mail: [email protected]
J. Boeckmans
e-mail: [email protected]
V. De Boe
Department of Urology, Universitair Ziekenhuis Brussel,
Laarbeeklaan 101, 1090 Brussels, Belgium
fatty liver disease (NAFLD) epidemic, which af￾fects about 24% of the global population
(Younossi et al. 2018). The prevalence of NASH
among NAFLD patients is projected to rise from
20% in 2015 to 27% by 2030 (Estes et al. 2018;
Chi 2015). The pathogenesis of NASH is multi￾factorial and poorly understood. Yet, it is clear that
the major driver of the outbreak of NASH is the
metabolic syndrome, in which insulin resistance
and abdominal obesity are key determinants
(Friedman et al. 2018). Nonetheless, the etiology
of NASH differs strongly among patients (Younossi
et al. 2018). A “multiple hit hypothesis” for the NASH
pathogenesis has been proposed, in which i.a.
lipotoxicity, insulin resistance, adipose tissue dysfunc￾tion, specific genetic polymorphisms, and epigenetic
factors are involved (Buzzetti et al. 2016).
Although weight loss is a prime and efficient
intervention in the treatment of NASH, it is often
hardly attainable. However, also, a number of pa￾tients are lean. In the past decade, tremendous
efforts have been made to develop an anti-NASH
therapy, yet, no drugs have been approved up to
date (Friedman et al. 2018). The market value is
impressive and is estimated to reach $15 billion by
2025 (Chi 2015).
One molecular strategy in anti-NASH drug de￾velopment relies on peroxisome proliferator￾activated receptor (PPAR) agonism. PPARs are nu￾clear receptors that regulate a multitude of process￾es associated with energy homeostasis and inflam￾mation (Gross et al. 2017). Consequently, PPARs
are attractive drug targets for metabolic syndrome￾related disorders. It has been demonstrated that
PPARA expression inversely correlates with the his￾tological severity of NASH (Francque et al. 2015)
and that PPAR-δ and PPAR-γ agonism attenuates
hepatic steatosis (Tong et al. 2019) and improves
insulin sensitivity (Heikkinen et al. 2007), respec￾tively. One of the furthest developed compounds is
elafibranor, a first-in-class dual PPAR-α/δ agonist
that is currently under clinical phase III evaluation
(Ratziu et al. 2016). Fu rthe rmo re, PPAR -α
(fibrates) and PPAR-γ (thiazolidinediones) agonists
proved earlier to be effective for the management of
dyslipidemia and type 2 diabetes, respectively. In
addition, pioglitazone, which is a thiazolidinedione,
is recommended for type 2 diabetes patients with
biopsy-proven NASH by the European Association
for the Study of the Liver (EASL) (Marchesini
et al. 2016) and the American Association for the
Study of Liver Diseases (AASLD) (Chalasani et al.
Today, animal models to investigate PPAR￾oriented anti-NASH effects fall behind due to the
high inter-species differences in PPAR functionality.
PPAR-α is, for example, much higher expressed in
rodent liver than in human liver. In addition, syn￾thetic PPAR-α agonists provoke peroxisome prolif￾eration and cancer in murine models (Holden and
Tugwood 1999). For these reasons, alternative
human-based hepatic cell systems are considered
more suitable to preclinically evaluate the efficacy
of novel anti-NASH compounds (Dash et al. 2017;
Boeckmans et al. 2018).
Recently, we established an “in vitro NASH model”
(Boeckmans et al. 2019) based on human skin precur￾sors (hSKP). hSKP are multipotent adult stem cells that
reside in the dermis and can be differentiated into hepa￾tocyte progenitor cells (hSKP-HPC) (Rodrigues et al.
2014). Upon triggering with NASH-inducing factors,
the hSKP-based model has been shown to correlate
reasonably with human NASH pathology. As such, it
could be used to unveil the anti-NASH properties of
elafibranor (Boeckmans et al. 2019). In the present
study, we aim at testing and classifying a series of PPAR
agonists (Table 1), namely bezafibrate (Franko et al.
2017) and fenofibrate (Fernández-Miranda et al. 2008)
(fibrates), pioglitazone (Sanyal et al. 2010) and
rosiglitazone (Ratziu et al. 2008) (thiazolidinediones),
lanifibranor (a first-in-class clinical phase II PPAR-pan
agonist) (Wettstein et al. 2017), saroglitazar (a clinical
phase II PPAR-α/γ agonist) (Jain et al. 2018),
pema fib rate (a selective PPAR -α modulator
(SPPARMα)) (Honda et al. 2017), and elafibranor
(Ratziu et al. 2016). Not only hSKP-HPC are used, but
also HepG2-, HepaRG-, and PHH-based in vitro
models of NASH. TGF-β-stimulated LX-2 stellate
cell cultures are also investigated in order to eval￾uate the anti-fibrotic properties of PPAR agonists.
First the ability of the different hepatic in vitro
models to reproduce molecular NASH characteris￾tics in vitro is investigated. Subsequently, the anti￾NASH potency of the aforementioned PPAR ago￾nists is assessed by investigating how they reverse
the NASH-specific cellular responses. Ultimately, a
combination of the most performant in vitro
models is used to grade the PPAR agonists by
their efficacy in vitro.
Materials and methods
In vitro hepatic “NASH” models
hSKP-HPC culture
After obtaining informed consent from the parents and
approval from the medical ethical committee of the UZ
Brussel, hSKP are isolated from foreskin of 2–10-year￾old boys and cryopreserved. hSKP are thawed at pas￾sage 6 and differentiated at passage 9 towards hSKP￾HPC as previously described (Rodrigues et al. 2014).
“NASH” conditions are mimicked as previously docu￾mented (Boeckmans et al. 2019). Briefly, the cells are
exposed for 24 h to fatty acids (65 μM sodium oleate
and 45 μM palmitic acid), 100 nM insulin, 4.5 mg/mL
glucose (all Sigma-Aldrich) and inflammatory cyto￾kines (50 ng/mL tumor necrosis factor (TNF)-α
(Prospec), 25 ng/mL interleukin (IL)-1β, and 8 ng/mL
transforming growth factor (TGF)-β (both Peprotech)).
HepG2 culture
HepG2 cells are cultured in 24- and 96-multiwell plates
(both Falcon BD) in Dulbecco’s modified Eagle’s medi￾um (DMEM) enriched with 4.5 mg/mL glucose and
glutamine (Lonza) supplemented with 10% (v/v) fetal
bovine serum (FBS) (Hyclone) and 1% (v/v) PenStrep
(Thermo Fisher Scientific). After thawing, the cells were
split twice using TrypLE Express (Thermo Fisher
Scientific). Exposures to NASH-triggers and PPAR ago￾nists were performed using DMEM containing 1.0 mg/
mL glucose (Lonza) supplemented with 10% (v/v) FBS,
1% (v/v) PenStrep (Thermo Fisher Scientific), and 2 mM
L-glutamine (Sigma-Aldrich). The “NASH” triggers are
identical to those described under ‘hSKP-HPC culture’
and are concomitantly added with the PPAR agonists.
HepaRG culture
Differentiated HepaRG cells with passage numbers be￾tween 13 and 20 (Biopredic International) are, after
thawing, cultured for seven days onto rat tail type I
collagen (Corning)-coated 24- and 96-multiwell plates
(both Falcon BD) according to the suppliers’ protocol.
Exposures to NASH triggers and PPAR agonists are
performed in William’s E medium (Thermo Fisher Sci￾entific) containing 10% (v/v) FBS (Hyclone), 1% (v/v)
PenStrep (Thermo Fisher Scientific), 2 mM L-glutamine
(Sigma-Aldrich), 872.69 nM insulin (Sigma-Aldrich),
and 0.5 nM hydrocortisone. “NASH” triggers are identi￾cal to those described under ‘hSKP-HPC culture’, except
for the insulin concentration that is raised to 8726.92 nM
due to the presence of insulin in the basal culture medium
and are concomitantly added with the PPAR agonists.
PHH culture
PHH cultures are delivered by Biopredic International.
The cells have been isolated and plated freshly onto
collagen-coated 24- and 96-multiwell plates. The donor
is free from human immunodeficiency viruses 1 and 2 and
hepatitis viruses B and C. After isolation, the cells are
maintained in long-term culture medium (Biopredic Inter￾national) and shipped at 37 °C. Four days after isolation,
Table 1 PPAR agonists and their PPAR isotype targets, phases of development relevant for NAFLD, identifiers (NCTs),
and (tentative) indications
PPAR agonist PPAR isotype Phase of development NCT (Tentative) indication
Bezafibrate α (/γ/δ) IV / Hyperlipidemia
Elafibranor α/δ III 02704403 NASH
Fenofibrate α IV / Hyperlipidemia
Lanifibranor α/γ/δ II 03008070 NASH
Pemafibrate α II 03350165 Dyslipidemia/NAFLD
Pioglitazone γ III/IV 00063622 Insulin resistance/diabetes type II/NASH
Rosiglitazone γ IV/withdrawn 01406704 Insulin resistance/diabetes type II
Saroglitazar α II 03061721 NASH
Cell Biol Toxicol
the cells are exposed for 24 h to conditions identical to
those described for HepaRG cell cultures.
LX-2 stellate cell culture
LX-2 cells are thawed at passage 9 in DMEM enriched
with 4.5 mg/mL glucose and glutamine (Lonza) contain￾ing 10% (v/v) fetal bovine serum (FBS) (Hyclone) and
1% (v/v) PenStrep (Thermo Fisher Scientific). The cells
are centrifuged for 3 min at 300 g and resuspended in the
same medium. Hereafter, the cells are cultured in 2% (v/v)
FBS medium and split at reaching 80% confluence at a
1:3 ratio using TrypLE reagent (Thermo Fischer Scien￾tific). The cells are split twice before conducting the
experiment. TGF-β is dissolved in PBS at a concentra￾tion of 5 μg/mL and used at a final concentration of
10 ng/mL. The cells are exposed for 24 h, and PPAR
agonists are concomitantly added with TGF-β.
Bezafibrate (Cayman Chemical Company), elafibranor
(Adooq Bioscience), fenofibrate, pioglitazone,
rosiglitazone (Sigma-Aldrich), lanifibranor, pemafibrate,
and saroglitazar (MedChemExpress) are dissolved in
dimethylsulfoxide (DMSO) (Sigma-Aldrich) in a 1000-
fold (200–100–60–30–10 mM) concentration according
to the final exposure concentration. The compounds are
added at sub-cytotoxic concentrations to the cultures for
24 h at the same time as the NASH triggers.
RNA extraction and RT-qPCR
After 24 h exposure, RNA is harvested from the 24-
multiwell plate wells (BD Falcon) using RNA lysis buffer
containing 1% v/v β-mercaptoethanol (Sigma-Aldrich).
Total RNA is purified using the GenElute Mammalian
Total RNA Purification Miniprep Kit (Sigma-Aldrich)
and quantified by a Nanodrop spectrophotometer (Thermo
Fisher Scientific). The iScript cDNA Synthesis Kit (Bio￾Rad Laboratories) is used for the generation of copy DNA,
which is purified using the GenElute PCR Clean-up Kit
(Sigma-Aldrich). A StepOne Plus system (Thermo Fisher
Scientific) is used for reverse transcription quantitative
polymerase chain reaction (RT-qPCR) using TaqMan
Mastermix and primers. Normalization of data is done
against the geometric mean of ubiquitin C and β2-
microglobulin (hSKP-HPC, HepG2, HepaRG, and PHH)
and glyceraldehyde 3-phosphate dehydrogenase and β2-
microglobulin (LX-2) using qBase+ software (Biogazelle).
Staining for neutral lipids
Cells are fixed in 24-multiwell plates with 4% (w/v) para￾formaldehyde in PBS for 10 min. BODIPY™ 493/503
staining for neutral lipids (Thermo Fisher Scientific) is
performed as previously documented (Boeckmans et al.
2019). Nuclei are stained using VECTASHIELD™
Mounting Medium containing DAPI (Vector Laborato￾ries). Fluorescence micrographs were taken using a
Nikon Eclipse Ti-S and quantified using ImageJ (n = 6).
The intracellular lipid load is calculated as a function of
BODIPY™ 493/503 lipid dye area and intensity.
Flow cytometry
Flow cytometric determination of neutral lipids is per￾formed as previously documented (Boeckmans et al.
2019). Briefly, cells are harvested from 24-multiwell plate
wells, centrifuged, resuspended, and stained with
BODIPY™ 493/503 neutral lipid stain (1:2500) and
Hoechst 33342 350/461 (1:1000) (both Thermo Fisher
Scientific). Analyses are performed using the Attune®
Acoustic Focusing Cytometer (Life Technologies) set to
a flow rate of 500 μL/min. Each run records 100,000
CellTiter-Glo® luminescent cell viability assay
The CellTiter-Glo® luminescent cell viability assay
(Promega) is used for the determination of intra￾cellular ATP levels. Cells are cultured in 96-
multiwell plates (Falcon BD). A total of 100 μL
CellTiter-Glo® reagent is added to each well, and
the plate is incubated for 10 min at room temper￾ature. A total of 150 μL per well is transferred to
a white 96-multiwell plate (Greiner Bio-One), and
luminescence signals are measured using a plate
reader (PerkinElmer).
Caspase-Glo® 3/7 assay
The Caspase-Glo® 3/7 assay (Promega) is used for the
determination of caspase-3/7 activity. Cells are cultured
in 96-multiwell plates (Falcon BD). A total of 100 μL
Caspase-Glo® 3/7 reagent is added to each well, and the
plate is incubated for 30 min at room temperature. A
Cell Biol Toxicol
total of 150 μL per well is transferred to a white 96-
multiwell plate (Greiner Bio-One), and luminescence
signals are measured using a plate reader (PerkinElmer).
Enzyme-linked immunosorbent assay
C-C motif chemokine ligand (CCL)2 and C-X-C
motif chemokine ligand (CXCL)5 enzyme-linked
immunosorbent assays (ELISAs) are purchased from
R&D systems and performed following the sup￾pliers’ instructions. Briefly, 200 μL assay diluent
RD1W is added to each well, followed by 50 μL
standard, blank, or sample at appropriate dilution
and incubated for 2 h at room temperature. Then,
the plate is washed three times using 400 μL wash
buffer per well and blotted against paper toweling.
Subsequently, 200 μL CCL2 or CXCL5 conjugate is
added to each well for 1 and 2 h, respectively,
followed by three washing steps. Hereafter, 200 μL
substrate solution is added to each well and incubat￾ed for 30 min in the dark. A total of 50 μL stop
solution is added, and the samples are homogenized
using a multichannel pipette. Signals are measured
at 450 nm, and wavelength correction is done at
550 nm using a plate reader (FLUOstar OPTIMA,
BMG Labtech). The chemokine secretions are nor￾malized against total ATP contents from the control
hSKP-HPC, HepG2, and HepaRG microarray data (n =
3) are obtained from Rodrigues et al. (2016b)
(GSE74000). PHH microarray data (n = 3) are obtained
from Thomas et al. (2012) (GSE31193). Patient datasets
are obtained from Frades et al. (2015) (n = 7 for healthy
liver; n = 9 for “NASH” liver) (GSE63068), Moylan
et al. (2014) (n = 32 for “severe NAFLD”)
(GSE31803), and Haas et al. (2019) (n = 13 for “no
NASH”; n = 57 for “baseline NASH”) (GSE106737).
The “healthy liver” datasets of Frades et al. (2015) are
used as controls for the “severe NAFLD” datasets of
Moylan et al. (2014). Data are normalized using Robust
Multichip Average (median polish).
Statistical analyses
GraphPad Prism 7.0 is used for statistical analyses. Data
are shown as the mean ± standard deviation, except for
flow-cytometric data that are shown as median ± stan￾dard deviation. A Student’s t test (when comparing two
groups) or a one-way ANOVA with post-hoc Sidak’s
multiple comparisons test (when comparing selected
pairs) is used. A minimum of three biological replicates
is used unless stated otherwise.
hSKP-HPC, HepG2, HepaRG, and PHH triggered
by lipogenic and inflammatory factors acquire
NASH-specific characteristics
In order to induce characteristics of NASH in vitro,
hSKP-HPC, HepG2, HepaRG, and PHH cultures
(Fig. 1a) are exposed for 24 h to lipogenic (glucose,
insulin, fatty acids) and inflammatory (TNF-α, IL-
1β, and TGF-β) triggers. All hepatic models show a
significantly increase in intracellular lipid load of at
least 2-fold relative to the untriggered controls. This
relative increase is the highest in hSKP-HPC follow￾ed by HepG2 and HepaRG cells and is the lowest in
PHH (Fig. 1b).
Increased expression of stearoyl-CoA desaturase-1
(SCD1), elongation of long-chain fatty acids 6
(ELOVL6), and carnitine palmitoyltransferase 1A
(CPT1A) is observed in NASH-triggered hSKP-HPC,
but not in the HepG2, HepaRG, and PHH models. On
the contrary, β-oxidative genes CPT1A and acyl-CoA
dehydrogenase short/branched chain (ACADSB) are
down-regulated in the HepG2, HepaRG, and PHH
models, which indicates that lipids could accumulate
through different mechanisms in distinct hepatic
in vitro models. Furthermore, the expression of cluster
of differentiation (CD)36 is potently down-regulated in
all hepatic in vitro “NASH” models (Fig. 1c). Gene
expression data amended to PHH values are shown in
Supplementary Fig. 1a. The expression of lipogenic
genes SCD1 and ELOVL6 is also increased in liver
samples of NAFLD patients (Supplementary Fig. 2a).
Furthermore, the expression of genes involved in β-
oxidation is also found to be decreased in the same
patient samples.
The expression of CD36 is found to be in￾creased in human NASH liver samples. However,
its downregulation has been reported in response
to TGF-β via a PPAR-γ dependent mechanism in
macrophages (Han et al. 2000) and its repression
Cell Biol Toxicol
could perhaps also result from a negative feedback
mechanism. Although these results do not per se
indicate that lipids accumulate through completely
distinct molecular mechanisms in the different
models, this suggests that different in vitro models
are required to correctly recapitulate different dis￾ease characteristics present in the liver of NASH
Analysis of the secretion of chemokines involved in the
hepatic recruitment of neutrophils and monocytes shows
Fig. 1 PHH, hSKP-HPC, HepG2, and HepaRG exhibit NASH
characteristics upon triggering by lipogenic and inflammatory
factors. a Phase-contrast images of hepatic in vitro systems. b
Neutral lipid (BODIPY™ 493/503 (green)) and nuclear staining
(DAPI (blue)) of hepatic in vitro models exposed to lipogenic and
inflammatory triggers. c Normalized mRNA levels (RT-qPCR) of
lipid metabolism-related genes in NASH-triggered hepatic in vitro
models. d Secretion (ELISA) and normalized mRNA levels (RT￾qPCR) of inflammatory chemokines in NASH-triggered hepatic
in vitro models. e ATP content and caspase-3/7 activity determi￾nation of NASH-triggered hepatic in vitro models. [Student’s t test
(*, **, and ***; p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001, respectively)]
Cell Biol Toxicol
an increase in CCL2 and CXCL5 in the NASH-triggered
in vitro systems, except for the secretion of CCL2 in
NASH-triggered HepG2 cells. This observation is con￾firmed at the mRNA level, for which besides a strong
upregulation of CCL2 (except in HepG2) and CXCL5,
also a drastic increase of CCL5 is measured (Fig. 1d).
These molecular alterations are also observed in NASH
human liver biopsies (Supplementary Fig. 2b).
Additionally, assessment of total energy levels shows
only an increased ATP-content in the NASH-triggered
hSKP-HPC model. On the contrary, increased caspase-
3/7 activity is observed in the NASH-triggered HepG2,
HepaRG, and PHH models, but not in the hSKP-HPC
cells (Fig. 1e). The increased caspase-3/7 activity in
PHH is, however, minor in comparison to the drastic
increase in NASH-triggered HepG2 and HepaRG cul￾tures. Non-normalized levels of cellular ATP content
and caspase-3/7 activity in the control conditions show
that PHH are not apoptotic before conducting the ex￾periments (Supplementary Fig. 3).
Fig. 2 Nuclear receptors and transcription factors related to lipid
metabolism and inflammation are dysregulated in NASH￾triggered human hepatic cultures. a Normalized mRNA levels
(microarray) of PPARs in untriggered hepatic in vitro models. b
Normalized mRNA levels (microarray) of PPAR co-activators in
untriggered hepatic in vitro models. c Normalized mRNA levels
(microarray) of PPAR co-repressors in untriggered hepatic in vitro
models. d Normalized mRNA levels (RT-qPCR) of PPAR isotypes
in NASH-triggered hepatic cultures. e Normalized mRNA levels
(RT-qPCR) of NFKB1 and RELA in NASH-triggered hepatic
cultures. f Normalized mRNA levels (microarray) of key inflam￾matory and lipogenic genes in untriggered hepatic in vitro models.
[One-way ANOVA with post-hoc Sidak’s multiple comparisons
test (when comparing selected pairs within more than 2 condi￾tions) or Student’st test (when comparing 2 conditions) (*, **, and
***; p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001, respectively)]
Cell Biol Toxicol
Lipid metabolism- and inflammation-related nuclear
receptors and transcription factors are dysregulated
in hepatic in vitro models of NASH
The expression of nuclear receptors and transcription
factors in NASH-triggered hepatic in vitro systems is
investigated to evaluate the impact of lipogenic and
inflammatory stimuli on upstream regulators of lipid
metabolism and inflammation.
Investigation of PPAR expression in untriggered
hSKP-HPC, HepG2, HepaRG, and PHH models shows
a more prominent expression of PPARA in HepaRG and
PHH cells when compared with hSKP-HPC and
HepG2. In contrast, PPARD and PPARG are higher
expressed in hSKP-HPC, HepG2, and HepaRG in com￾parison to PHH cells (Fig. 2a). Also the expressions of
PPAR co-activators (Fig. 2b) are different among the
in vitro systems. The expressions of E1A Binding Pro￾tein P300 (EP300) and Nuclear Receptor Coactivator 1
(NCOA1) are significantly higher in hSKP-HPC in com￾parison to the other in vitro systems while hSKP-HPC
show the lowest expression of peroxisome proliferator￾activated receptor-gamma coactivator 1 alpha
(PPARGC1A). On the contrary, PHH and HepaRG ex￾hibit the lowest expression of PPAR co-repressors Nu￾clear Receptor Interacting Protein 1 (NRIP1) and Nu￾clear Receptor Corepressor 2 (NCOR2) in comparison
to the hSKP-HPC and HepG2 models (Fig. 2c).
PPARA is decreased in NASH-triggered HepaRG and
PHH cultures (Fig. 2d) as earlier reported (Francque et al.
2015) and also observed in public microarray data of
NASH liver biopsies (Supplementary Fig. 2c). NASH￾triggered hSKP-HPC and PHH cultures exhibit increased
PPARD expression. PPARG is down-regulated in HepG2
and HepaRG models, slightly increased in hSKP-HPC,
and unaltered in the PHH “NASH” cultures.
As sustained inflammatory conditions can lead to de
novo synthesis of inflammatory transcription factors
(Bierhaus et al. 2001; Dorn et al. 2014); nuclear factor
kappa B P65 (RELA) and nuclear factor kappa B sub￾unit 1 (NFKB1) expression has also been investigated.
With the exception of HepG2 cells, all NASH￾triggered cultures show a prominent transcriptional
induction of NFKB1. This is, although much less
prominent, also observed for RELA (Fig. 2e) in
case of hSKP-HPC and PHH. This can be ex￾plained by the fact that the p50 promotor can be
activated by both p50 and p65 subunits, and more
strongly by the combination of both (Ten et al.
1992), suggesting an important contribution of
NF-κB signaling to the inflammatory response.
Gene expression data of PPARs, NFKB1, and RELA
amended to PHH values are shown in Supplemen￾tary Fig. 4.
The differences observed among the different
“NASH” in vitro models could be explained by the basal
distinct expression of lipogenic and inflammatory genes
(Fig. 2f). HepG2, for example, poorly expresses inter￾leukin 1 receptor type 1 (IL1R1), which is required for
IL-1β-induced inflammation. Apart from HepG2, also
HepaRG and PHH express TNF-α, IL-1β, and TGF-β
receptors at lower levels than hSKP-HPC do. Genes
related to lipid metabolism (i.e., lipid transport, de novo
lipogenesis, and β-oxidation) are, however, generally
higher expressed in PHH cultures.
PPAR agonists differently reduce lipid loads in distinct
NASH-triggered cultures
NASH-triggered hSKP-HPC, HepG2, HepaRG, and
PHH cultures are exposed for 24 h to bezafibrate,
elafibranor, fenofibrate, lanifibranor, pemafibrate,
pioglitazone, rosiglitazone, and saroglitazar at
60 μM at which no cytotoxicity occurs (Supple￾mentary Fig. 5). Figure 3a shows fluorescence
micrographs of NASH-triggered cultures exposed
to PPAR agonists, stained for neutral lipids. Quan￾tification of these responses reveals that pioglita￾zone and saroglitazar reduce the lipid load in all
tested in vitro NASH models, with the exception
of HepaRG, for which no alterations in lipid load
are observed for any of the compounds tested (Fig.
3). Pemafibrate reduces lipid levels only in HepG2
and PHH, while elafibranor and rosiglitazone re￾duce the amount of lipids only in hSKP-HPC and
PHH. Fenofibrate, however, further increases lipid
accumulation in hSKP-HPC.
To further investigate the effects of PPAR agonists
on lipid accumulation, measurement of intracellular
lipids was performed by flow cytometry using hSKP￾HPC cells. These cells, together with PHH, show the
biggest modulation in neutral lipid load upon expo￾sure to PPAR agonists, yet the employability of PHH
for flow cytometry is technically unfeasible due to
the fragility of the cells. Conversely, steatotic hSKP￾HPC have previously been successfully analyzed by
flow cytometry (Rodrigues et al. 2016a). The histo￾gram in Supplementary Fig. 6a shows a distinct shift
Cell Biol Toxicol
to the right when the cells are triggered by “NASH”
factors, and a shift to the left when PPAR agonists are
added, indicating a reduction in intracellular lipid
levels and confirming the imaging results (Supple￾mentary Fig. 6b). As previously observed,
fenofibrate induces the opposite effect by shifting
the histogram to the right, indicating an even higher
lipid load than untreated “hSKP-HPC NASH” cells
(Supplementary Fig. 6a).
The combination of the lipid readout with the side￾scatter (SSC), revealing the cellular granularity density
plots, allows for the differentiation of cells representing
“LipidLow,” “LipidHigh,” and “no NAFLD” for each
individual condition (Supplementary Fig. 6c). “No
NAFLD” thresholds are defined by the hSKP-HPC control
condition that falls for more than 99% in the bottom-left
quadrant. “LipidLow” is defined by the lower SCC com￾bined with an increased fluorescence lipid signal.
“LipidHigh” holds, apart from increased lipid dye signal,
also a higher SSC, indicating a higher granularity due to
the abundantly present lipids. The top-left quadrant corre￾sponds to cytotoxicity or other cellular events that increase
cell granularity, with no effect on the lipid load. When
examining the percentage of each sample in the
“LipidLow” and “LipidHigh” groups, a distinction in
anti-steatotic effects can be made. More than 60% of each
treated sample, with the exception of fenofibrate-treated
samples, falls within the “no NAFLD” group, indicating
that they all have anti-steatotic properties. Elafibranor- and
rosiglitazone-treated samples show the highest effect, ac￾counting for 90% in the “no NAFLD” group and less than
5% in each of the “LipidLow” and “LipidHigh” groups.
Using this assay, repartitioning of lipids is observed for
pemafibrate and lanifibranor that reduce “LipidLow” cells
while increasing “LipidHigh” cells. Furthermore,
saroglitazar potently reduces “LipidLow” cells, while cells
that hold more lipid droplets remain unaffected, indicating
subtle different anti-steatotic properties among PPAR
Although we here investigate whether PPAR agonists
are able to block lipid accumulation, this approach can
also be applied for reversing the steatotic phenotype. For
example, hSKP-HPC exposed to “NASH” triggers for
24 h followed by elafibranor treatment at either 30 or
Fig. 3 PPAR agonists reduce lipid loads in human in vitro models
of NASH. a Micrographs of in vitro NASH models stained for
neutral lipids with BODIPY™ 493/503 (green) and DAPI (blue,
nuclei). b Quantification of neutral lipid load (area × intensity).
[One-way ANOVA with post-hoc Sidak’s multiple comparisons
test vs control samples (***; p ≤ 0.001) and vs “NASH” ($, $$,
and $$$; p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001, respectively)]
Cell Biol Toxicol
60 μM for an additional 24 h show reduced lipids when
treated with the latter concentration (Supplementary
Fig. 7). Notably, earlier research showed that using a
co-exposure approach, also a concentration of 30 μM
elafibranor can block lipid accumulation (Boeckmans
et al. 2019), suggesting this experimental design could
be better suited for high-throughput screening.
PPAR agonist-mediated lipid reduction is acquired
through different modes of action depending
on the hepatic in vitro model
Figure 4 shows a heatmap of the regulation of genes
involved in lipid metabolism by PPAR agonists com￾pared with the untreated NASH-triggered cultures.
Markers of lipogenesis, i.e., SCD1 and ELOVL6, are
significantly downregulated by elafibranor in hSKP￾HPC and HepG2 cells. Contradictory, other PPAR ago￾nists induce SCD1 and ELOVL6 as well as CPT1A,
which regulates the transport of fatty acids into the
mitochondria for β-oxidation, in NASH-triggered
HepG2, HepaRG, and PHH cultures. Yet, no significant
increase in ACADSB expression is observed in any
model. The expression of CD36 is induced by
fenofibrate in hSKP-HPC, while elafibranor,
lanifibranor, pemafibrate, rosiglitazone, and saroglitazar
potently induce CD36 in HepaRG “NASH” cultures. In
NASH-triggered PHH cultures, increased CD36 expres￾sion is seen upon exposure of the cells to lanifibranor,
pemafibrate, pioglitazone, and rosiglitazone.
As such, underlying molecular mechanisms that lead
to reduction of lipids induced by PPAR agonists can
vary based on the model that is used, which indicates the
need of multiple models for studying (novel) drugs.
Gene expression data amended to PHH values are
shown in Supplementary Fig. 8, in which the white color
indicates expression values of untriggered PHH.
PPAR agonists differently reduce the secretion
and expression of inflammatory chemokines
The modulation in expression of chemokines that play a
role in the recruitment of monocytes and neutrophils to
the liver shows indirectly the potential of PPAR agonists
Fig. 4 Anti-steatotic modes of action of PPAR agonists differ
between hepatic in vitro systems. Modulation of normalized
mRNA levels (RT-qPCR) of lipid metabolism-related genes by
PPAR agonists in in vitro NASH models. [One-way ANOVA with
post-hoc Sidak’s multiple comparisons test vs “NASH” (*; p ≤
Cell Biol Toxicol
to induce an anti-inflammatory response. It was ob￾served that secretion of both CCL2 and CXCL5 is
reduced by PPAR agonists. Yet, important differences
are observed in the potency by which the compounds
reduce the chemokine levels.
Different in vitro models also show different se￾cretion profiles of these cytokines. Pioglitazone re￾duces CXCL5 levels in hSKP-HPC, HepaRG, and
PHH, but not in HepG2 cultures. On the contrary,
pioglitazone reduces CCL2 secretion in HepaRG
and PHH, but not in hSKP-HPC cultures (Fig. 5a).
Furthermore, elafibranor tremendously lowers CCL2
and CXCL5 levels in hSKP-HPC, while this is only
moderate in PHH and absent in HepaRG for CCL2.
The alterations in chemokine secretion correspond
well to the modulation of gene expression. In addi￾tion, the strongest reductions in chemokines, in￾duced by different PPAR agonists, correspond to
the significant reduction in expression of NFKB1,
as their upstream regulator (Fig. 5b).
PPAR agonists restrict pro-fibrotic gene expression
Liver fibrosis is an important predictor of NAFLD￾related mortality (Ekstedt et al. 2015). To evaluate the
anti-fibrotic properties of PPAR agonists, LX-2 stellate
cells are exposed to TGF-β and the compounds under
evaluation. Although these cells exhibit an activated
phenotype (Xu et al. 2005), exposure to TGF-β at a
concentration of 10 ng/mL for 24 h further induces actin
alpha 2, smooth muscle (ACTA2) and collagen type I
alpha 1 (COL1A1) gene expression, which are promi￾nent pro-fibrotic genes. Yet, this cannot be observed for
lysyl oxidase like 2 (LOXL2) (Fig. 6a), which intervenes
in the cross-linking of collagens. Exposure of LX-2
cultures to TGF-β at the tested concentration does not
affect cell viability (Supplementary Fig. 9a). Sub￾cytotoxic concentrations of PPAR agonists are used in
the exposure experiments (Supplementary Fig. 9b).
Bezafibrate, pioglitazone, rosiglitazone, and saroglitazar
are used at 60 μM, elafibranor and fenofibrate at 30 μM,
and lanifibranor and pemafibrate at 10 μM. At these
concentrations, it is observed that elafibranor,
fenofibrate, pioglitazone, and saroglitazar significantly
reduce ACTA2 levels. Moreover, elafibranor also re￾presses COL1A1 and LOXL2 mRNA levels (Fig. 6b),
emphasizing the potential anti-fibrotic properties of this
In vitro scoring of anti-NASH efficacy unveils
anti-NASH potency of PPAR agonists
In clinical studies, the NAFLD activity score (NAS) is
used to grade NASH biopsies and consists of the
evaluation of (a) steatosis, (b) lobular inflammation,
and (c) hepatocyte ballooning (Kleiner et al. 2005). In
analogy with this classification, we developed an
in vitro scoring system to classify compounds accord￾ing to their anti-NASH potency. Steatosis and inflam￾mation are evaluated in vitro (Fig. 7a) by means of the
percentages of intracellular lipid load and CCL2 and
CXCL5 secretions, respectively, relative to the NASH￾triggered cultures. To cover the whole spectrum of
NASH, and in analogy with the Steatosis, Activity
(ballooning + lobular inflammation), and Fibrosis
(SAF) score (Bedossa et al. 2012), hepatocellular bal￾looning, which is the third feature of NAS, is substitut￾ed in vitro by ACTA2 and COL1A1 expressions in LX-
2 stellate cells, representing key genes involved in
fibrosis. Thresholds for the features of the in vitro
scoring system are based on NAS thresholds and are
displayed in Fig. 7a.
In line with NAS, a score higher or equal to 5 indi￾cates no or low anti-NASH potency, while a score below
3 refers to strong anti-NASH effects. Figure 7b shows
the application of the in vitro scoring system that differ￾entiates the anti-steatotic, anti-inflammatory, and anti￾fibrotic properties of PPAR agonists. When taking all
evaluated in vitro models into account (indicated as
“combi”’ in Fig. 7c) by making the mean of all obtained
scores, elafibranor is identified as having the highest
anti-NASH potency, while saroglitazar, pioglitazone,
and pemafibrate fall in the “intermediate” class. Yet,
considering that NASH-triggered HepaRG cultures do
not sensitively respond to PPAR agonists to attenuate
the accumulating lipids and that NASH-triggered
HepG2 cultures do not secrete CCL2 in this experimen￾tal context, a combination of in vitro data obtained from
the respective steatosis and inflammation readouts in
PHH and hSKP-HPC alone could be used for the
in vitro scoring system.
As such, when using only the in vitro data obtained in
PHH, hSKP-HPC, and LX-2 cultures, elafibranor is
classified as having “high anti-NASH potency,” while
saroglitazar, pioglitazone, rosiglitazone, and
pemafibrate are classified as having “moderate anti￾NASH potency” (Fig. 7c). These findings correspond
to current clinical data.
Cell Biol Toxicol
With the global rising in NASH prevalence, the need for
effective anti-NASH therapies has become urgent. Multiple
clinical phase II and III trials are currently ongoing, but up to
now, no anti-NASH drug has been approved (Friedman
et al. 2018). Moreover, discontinuation of clinical trials and
midterm reports of running studies have shown no or only
limited efficacy of several anti-NASH compounds under
development (Friedman et al. 2018; Boeckmans et al.
Fig. 5 PPAR agonists distinctly attenuate mediators of inflamma￾tion in in vitro models of NASH. a Modulation of CCL2 and
CXCL5 secretions (ELISA) by PPAR agonists in in vitro NASH
models. b Spider plots in which the outer borders represent the
normalized mRNA levels (RT-qPCR) of CCL2, CXCL5, CCL5,
and NFKB1 in the NASH-triggered cultures and the middle points
represent the control conditions. [One-way ANOVA with post-hoc
Sidak’s multiple comparisons test vs control samples (***; p ≤
0.001) and vs “NASH” ($, $$, and $$$; p ≤ 0.05, p ≤ 0.01, and p ≤
0.001, respectively)]
Cell Biol Toxicol
2020). One particular class of anti-NASH drugs targets
different PPAR isotypes. Among this class are elafibranor
and saroglitazar, both promising drugs for anti-NASH ther￾apy (Ratziu et al. 2016; Gawrieh et al. oral communication,
“The Liver Meeting (AASLD)” 2019, Boston, USA).
The lack of predictive, human-relevant in vivo NASH
models early during the drug development process may
have contributed to these disappointing results. Hence,
in vitro methodologies that can accurately recapitulate
different molecular and cellular features of human NASH
are well positioned to advance anti-NASH drug develop￾ment (Boeckmans et al. 2018). In this context, Feaver
et al. established a dynamic NASH co-culture model
consisting of PHH, stellate cells, and macrophages ex￾posed to glucose, insulin, and fatty acids. Their system
predicted the anti-NASH effects of obeticholic acid, a
phase III anti-NASH drug (Feaver et al. 2016). Yet, the
use of such a sophisticated co-culture system of primary
cells is unpractical and not suitable for high-throughput
screening. Instead, we simulate the presence of activated
non-parenchymal cells and metabolic syndrome features
in vitro by supplementing the cell culture medium with
the most critical factors characteristic of NASH (insulin,
glucose, fatty acids, and inflammatory cytokines). Fol￾lowing this “in vitro NASH environment” approach, we
could previously demonstrate that human stem cell￾derived hSKP-HPC are able to recapitulate the anti￾NASH effects of elafibranor (Boeckmans et al. 2019).
In the present study, this set-up was further expanded to 3
of the most frequently used hepatic in vitro models, i.e.,
HepG2, HepaRG, and PHH. As such, we compared the
capacity of each hepatic cell system to reproduce specific
NASH characteristics in vitro to the hSKP-HPC model
and subsequently investigated in each cell model the
potential anti-NASH effects of multiple PPAR agonists
under clinical development.
Based on selected markers involved in de novo lipo￾genesis and β-oxidation, it seemed that HepG2-,
HepaRG-, and PHH-based models are the most suitable
to investigate NASH-related β-oxidation, while hSKP￾HPC exhibited the highest increase in genes (SCD1,
ELOVL6) related to de novo lipid synthesis.
hSKP-HPC exhibited the lowest basal expression of
PPARGC1A, which is a master regulator of mitochon￾drial biogenesis (Scarpulla 2011; LeBleu et al. 2014),
while it exhibited the highest TGFBR2 expression that is
involved in TGF-β-mediated regulation of de novo
lipogenesis and fatty acid β-oxidation in hepatocytes
(Yang et al. 2014). This could explain these fundamental
differences in behaviors towards “NASH” triggers
in vitro. Remarkably, all hepatic in vitro “NASH”
models exhibited decreased CD36 expression, while
this fatty acid transporter is often found to be upregulat￾ed in human NASH (Miquilena-Colina et al. 2011). Yet,
CD36 expression is reduced in NASH-related hepato￾cellular carcinoma (Kücükoglu et al. 2017), which sug￾gests that the cocktail of “NASH”-inducing factors that
was used in this study promotes the development of
an advanced disease state. This correlates with in￾creased CXCL5 secretion that was seen in all
in vitro NASH models and whereof hepatic expression
is increased in advanced fibrosis and cirrhosis (Tacke
et al. 2011).
Exposure to lipogenic and inflammatory triggers in￾duced lipid accumulation and secretion of chemokines
involved in neutrophil and monocyte recruitment in all
hepatic cell systems. NASH-triggered HepG2 cultures do
not express CCL2, which is secreted by injured hepato￾cytes and Kupffer cells to promote hepatic infiltration and
accumulation of CCR2-expressing monocytes from the
bone marrow. CCL2 is also implicated in insulin resistance
and steatosis development (Marra and Tacke 2014).
Fig. 6 PPAR agonists reduce pro-fibrotic gene expression levels
in human LX-2 stellate cells. a Normalized mRNA levels (RT￾qPCR) of pro-fibrotic genes in LX-2 cells after stimulation with
TGF-β. b Modulation of normalized mRNA levels (RT-qPCR) of
pro-fibrotic genes by PPAR agonists in TGF-β-stimulated LX-2
cells. [a Student’s t test (*; p ≤ 0.05); b One-way ANOVA with
post-hoc Sidak’s multiple comparisons test vs TGF-β-treated sam￾ples ($, $$, and $$$; p ≤ 0.05, p ≤ 0.01, and p ≤ 0.001,
Cell Biol Toxicol
Although no induction in CCL2 gene expression nor
secretion could be observed in HepG2 cultures, this could
possibly be induced by supplementing this culture with IL-
1α (Ohashi et al. 2009) or lipopolysaccharide (Kanmani
Fig. 7 Scoring system to classify PPAR agonists according to
their in vitro anti-NASH potency. a Schematic representation of
the development of an “in vitro NAFLD activity score” for grading
of potential anti-NASH features of PPAR agonists. b In vitro
scoring system differentiates anti-steatotic, anti-inflammatory,
and anti-fibrotic properties of PPAR agonists. c Elafibranor,
followed by saroglitazar, pioglitazone, rosiglitazone, and
pemafibrate show the strongest anti-NASH effects based on a
combination of readouts from PHH, hSKP-HPC, and LX-2
Cell Biol Toxicol
and Kim 2018). Hepatocyte apoptosis is an important
feature of NASH (Tarek et al. 2011), which also occurred
in NASH-triggered HepG2, HepaRG, and PHH cultures
after 24 h incubation, while this was earlier observed in
hSKP-HPC after 72 h (Boeckmans et al. 2019).
NASH patients typically exhibit decreased liver
PPARA expression (Francque et al. 2015). This feature
is also present in NASH-triggered HepG2, HepaRG, and
PHH, indicating their relevance to the disease. This was
accompanied by a decreased CPT1A expression, which is
a PPAR-α target gene (Mandard et al. 2004). Expression
of NF-κB subunits is also increased in NASH-triggered
cultures, with the exception of HepG2. The increased
transcription of RELA and NFKB1 suggests the presence
of sustained inflammation which overrides the endoge￾nous negative feedback (Bierhaus et al. 2001). Further￾more, the p50 promotor can be activated by both p50 and
p65 subunits (Ten et al. 1992).
Eight PPAR agonists, in different phases of develop￾ment, were tested for their ability to reduce NASH char￾acteristics in vitro.
All compounds were tested at a concentration of 60 μM
among the different hepatic in vitro systems. Yet, one could
raise the question whether this concentration was not too
low for some compounds to have an effect. However,
when PPAR agonists are administrated to patients, peak
plasma concentrations of ± 14.0 μM (fenofibrate 200 mg)
and 3.5 μM (pioglitazone 45 mg) have been measured
(Vaidyanathan et al. 2008), which is below the concentra￾tion of 60 μM. Hence, testing higher compound concen￾trations would go along with decreasing human relevance.
Pioglitazone and saroglitazar, a PPAR-γ (Sanyal et al.
2010) and PPAR-α/γ (Jain et al. 2018) agonist, respective￾ly, reduced lipid loads in all models with the exception of
HepaRG. Indeed, HepaRG cultures were not able to cap￾ture potential anti-steatotic effects of PPAR agonists. In
contrast, multiple compounds induced CPT1A expression
in HepaRG, which has been previously observed during
NASH relieve in patients (Francque et al. 2015). Yet,
earlier research with HepaRG showed that among others,
rosiglitazone, fenofibrate, and bezafibrate reduce oleic
acid-induced intracellular lipids after 14 days. As such,
this could implicate that certain in vitro systems need
long-term repeated exposure to unveil anti-steatotic prop￾erties of (potential) drugs (Rogue et al. 2014), which,
however, hampers fast high-throughput screening.
Elafibranor, pemafibrate, and rosiglitazone additionally
reduce lipid accumulation in PHH and hSKP-HPC. Thus,
despite harboring differences in PPAR biology, alternative
cellular systems are capable of reproducing PHH-mediated
results for specific applications, which could potentially be
attributed to distinct expressions of PPAR co-activators
and/or co-repressors. Yet, modes of action of certain com￾pounds could differ depending on the model used. For
example, elafibranor potently reduces SCD1 and ELOVL6
expression in hSKP-HPC, while a strong induction of
CPT1A is observed in PHH. Intriguingly, several PPAR
agonists induce CD36 expression, which is a known
PPAR-γ target gene (Maréchal et al. 2018), in HepaRG
and PHH cultures. Yet, PPAR-α has been also identified as
a regulator of CD36 (Sato et al. 2002). Indeed, also
pemafibrate, a SPPARMα, induces CD36 expression in
NASH-triggered HepaRG and PHH cultures. This effect
seems ambiguous, since increased CD36 expression has
been linked to NASH (Miquilena-Colina et al. 2011), but
goes along with the decrease in lipid accumulation in the
present study. Nevertheless, hepatic CD36 overexpression
has been also related to ameliorating insulin resistance,
glycogen homeostasis, and attenuation of hepatic steatosis
in mice, corroborating current data (Garbacz et al. 2016).
Furthermore, earlier research with rosiglitazone,
troglitazone, muraglitazar, and tesaglitazar showed that
these PPAR agonists induce CD36 expression in HepaRG
and PHH cultures (Rogue et al. 2011).
PPAR agonists reduce chemokine secretions to different
extents which indicates distinctive anti-inflammatory prop￾erties. Yet, also major intra-compound differences are ob￾served between the models. For example, CXCL5 is the
most strongly reduced in HepaRG and PHH by
lanifibranor, while elafibranor completely restricts CXCL5
secretion in hSKP-HPC and HepG2 that only show mod￾erate effects for lanifibranor. Furthermore, HepaRG cells
only show reduced CCL2 secretion for pioglitazone, while
other PPAR agonists clearly also reduce CCL2 levels in
The development of liver fibrosis is often associated
with NASH and is the best predictor of disease-specific
mortality. As such, it better predicts overall mortality than
the often-used NAS (Ekstedt et al. 2015). Therefore, the
potential anti-fibrotic effects of PPAR agonists were inves￾tigated as well. Elafibranor, fenofibrate, pioglitazone, and
saroglitazar potently restricted ACTA2 expression, which is
a classical marker for stellate cell activation (Mannaerts
et al. 2015), in TGF-β-exposed LX-2 cells. Elafibranor
additionally reduced COL1A1 and LOXL2 which indicates
additional anti-fibrotic properties.
In order to further differentiate the potential anti-NASH
efficacies of PPAR agonists, we have developed an in vitro
Cell Biol Toxicol
scoring system related to NAS and SAF score, which are
used for the follow-up of NASH patients during clinical
trials (Brunt et al. 2011). NAS consists of the following
three scoring features: (a) steatosis (0–3), (b) lobular in￾flammation (0–3), and (c) hepatocyte ballooning (0–2)
(Kleiner et al. 2005). Yet, since fibrosis is a major deter￾minant of mortality, our in vitro scoring system is based on
(a) steatosis (0–3), inflammation (0–3), and fibrosis (0–2)
(Ekstedt et al. 2015), which are the three pillars of NAFLD
progression, as also used in the SAF scoring system
(Bedossa et al. 2012). Since NASH-triggered HepaRG
and HepG2 cultures did not show sensitive lipid modula￾tion in response to PPAR agonists and did not secrete
CCL2, respectively, we evaluated if a selection of models
consisting of LX-2 stellate cells, PHH, and hSKP-HPC
alone could also be used to classify the compounds for
their anti-NASH efficacies. A combination of PHH and
hSKP-HPC was found to be sufficient to study de novo
lipogenesis and β-oxidation modifications induced by
PPAR agonists. Using PHH, hSKP-HPC, and LX-2, a
similar, but more stringent ranking is observed, which
could facilitate the implementation of such a scoring sys￾tem in drug development.
Using the in vitro scoring system, elafibranor and
saroglitazar, being respectively evaluated in a clinical phase
III and II trial, are identified as the most effective anti￾NASH compounds (in vitro NAS of 1 and 3, respectively).
This corresponds to present clinical data, in which
elafibranor showed to resolve NASH without worsening
of fibrosis in a post-hoc analysis (Ratziu et al. 2016). On the
other hand, it has recently been reported that saroglitazar
improves steatosis and reduces serum alanine aminotrans￾ferase (ALT) levels in NASH patients in a phase II study
(EVIDENCES IV) (Gawrieh et al. oral communication,
“The Liver Meeting (AASLD)” 2019, Boston, USA). Sec￾ond in line are pioglitazone, rosiglitazone, and pemafibrate.
Pioglitazone is currently recommended by the AASLD and
EASL for anti-NASH treatment in biopsy-proved NASH
patients suffering from type 2 diabetes (Marchesini et al.
2016; Chalasani et al. 2012; Sanyal et al. 2010). Clinical
trials for the evaluation of rosiglitazone for anti-NASH
treatment show important anti-steatogenic effects, yet with￾out other major anti-NASH features (Ratziu et al. 2008;
Ratziu et al. 2010). Interestingly, rosiglitazone potently
reduces lipid levels in the investigated cellular models of
NASH but has no important beneficial effects on inflam￾matory nor fibrotic parameters that are studied in vitro,
correlating to the clinical data. Rosiglitazone has been with￾drawn from the market due to cardiotoxicity (Varga et al.
2015) but clearly demonstrates the pressing need for appro￾priate in vitro models and the value of preclinical in vitro
evaluation. Pemafibrate, which is currently under phase II
evaluation for NAFLD, already showed to reduce ALT,
gamma glutamyl transferase (GGT), and triglyceride serum
levels in a clinical phase III trial for dyslipidemia, while
fenofibrate increased ALT and GGT levels (Ishibashi et al.
2018). Bezafibrate and fenofibrate have been found to be
ineffective for NASH (Sumida and Yoneda 2018). Further￾more, the steatosis-inducing properties of fenofibrate, as
observed in fenofibrate-treated NASH-triggered hSKP￾HPC cultures, have been previously reported as well in
hepatic in vitro models and NAFLD patients (López-Riera
et al. 2017; Yan et al. 2014). Lanifibranor, on the other hand,
has been shown to improve steatosis, inflammation, and
fibrosis in vivo, yet no clinical data of the current phase II
trial have been divulgated (Wettstein et al. 2017). Hence,
combining in vitro models could be key for unveiling anti￾NASH properties of novel compounds within a short
timeframe to identify clinically relevant compounds by
high-throughput screening.
Commonly used human hepatic in vitro models are capa￾ble of reproducing characteristics of the human NASH
pathology upon triggering with disease-relevant factors.
Hence, these in vitro hepatic “NASH” models are able to
identify anti-NASH characteristics of PPAR agonists.
Most stringent results in terms of phenotypic anti-NASH
responses are obtained using PHH- and hSKP-HPC￾derived models and a combination thereof. The in vitro
scoring system based on PHH, hSKP-HPC, and LX-2
stellate cells shows that elafibranor and saroglitazar,
followed by pioglitazone, hold the strongest anti-NASH
properties, which corresponds to available clinical data.
Author contributions Conception and design: J.B. and R.M.R;
data analysis and interpretation: J.B., J.D.K., and R.M.R.; collection
and assembly of data: J.B., A.N., M.R., K.B., B.C., and A.H.;
financial support: V.R., T.V., and R.M.R.; manuscript writing: J.B.,
A.N., M.R., K.B., B.C., A.H., V.R., J.D.K., T.V., and R.M.R.; admin￾istrative support: V.R. and T.V.; provision of study material: V.D.B.;
project supervision: V.R., T.V., and R.M.R.; final approval of manu￾script: J.B., T.V., and R.M.R.
Funding information This work was funded by grants of Re￾search Foundation Flanders (1S10518N, 12H2216N, 1S73019N, and
G042019N), Onderzoeksraad Vrije Universiteit Brussel, and the Re￾search Chair Mireille Aerens for Alternatives to Animal Testing.
Cell Biol Toxicol
Compliance with ethical standards
Conflict of interest The authors have no declarations of interest.
Ethics approval Ethical approval was obtained from the medi￾cal ethical committee of the UZ Brussel (Belgium).
Consent to participate Informed consent was obtained where
Consent for publication All authors have read and agreed to the
published version of the manuscript.
Bedossa P, Poitou C, Veyrie N, Bouillot JL, Basdevant A, Paradis
V, et al. Histopathological algorithm and scoring system for
evaluation of liver lesions in morbidly obese patients.
Hepatology. 2012;56:1751–9.
Bierhaus A, Schiekofer S, Schwaninger M, Andrassy M, Humpert
PM, Chen J, et al. Diabetes-associated sustained activation of
the transcription factor nuclear factor-κB. Diabetes. 2001;50:
Boeckmans J, Natale A, Buyl K, Rogiers V, De Kock J, Vanhaecke
T, et al. Human-based systems: mechanistic NASH model￾ling just around the corner? Pharmacol Res. 2018;134:257–
Boeckmans J, Buyl K, Natale A, Vandenbempt V, Branson S, De
Boe V, et al. Elafibranor restricts lipogenic and inflammatory
responses in a human skin stem cell-derived model of NASH.
Pharmacol Res. 2019;144:377–89.
Boeckmans J, Natale A, Rombaut M, Buyl K, Rogiers V, De Kock
J, et al. Anti-NASH drug development hitches a lift on PPAR
agonism. Cells. 2020;9:1–20.
Brunt EM, Kleiner DE, Wilson LA, Belt P, Neuschwander-Tetri
BA. The NAS and the histopathologic diagnosis of NAFLD:
distinct clinicopathologic meanings. Hepatology. 2011;53:
Buzzetti E, Pinzani M, Tsochatzis EA. The multiple-hit pathogen￾esis of non-alcoholic fatty liver disease (NAFLD).
Metabolism. 2016;65:1038–48.
Chalasani N, Younossi Z, Lavine JE, Diehl AM, Brunt EM, Cusi K,
et al. The diagnosis and management of non-alcoholic fatty
liver disease: practice guideline by the American Association
for the Study of Liver Diseases, American College of
Gastroenterology, and the American Gastroenterological
Association. Hepatology. 2012;55:2005–23.
Chi KR. The NASH drug dash. Nat Rev Drug Discov. 2015;14:
Dash A, Figler RA, Blackman BR, Marukian S, Collado MS,
Lawson MJ, et al. Pharmacotoxicology of clinically￾relevant concentrations of obeticholic acid in an organotypic
human hepatocyte system. Toxicol Vitr. 2017;39:93–103.
Dorn C, Engelmann JC, Saugspier M, Koch A, Hartmann A,
Müller M, et al. Increased expression of c-Jun in nonalcohol￾ic fatty liver disease. Lab Investig. 2014;94:394–408.
Ekstedt M, Hagström H, Nasr P, Fredrikson M, Stål P, Kechagias
S, et al. Fibrosis stage is the strongest predictor for disease￾specific mortality in NAFLD after up to 33 years of follow￾up. Hepatology. 2015;61:1547–54.
Estes C, Razavi H, Loomba R, Younossi Z, Sanyal AJ. Modeling
the epidemic of nonalcoholic fatty liver disease demonstrates
an exponential increase in burden of disease. Hepatology.
Feaver RE, Cole BK, Lawson MJ, Hoang SA, Marukian S,
Blackman BR, et al. Development of an in vitro human liver
system for interrogating nonalcoholic steatohepatitis. J Clin
Invest. 2016;1:e90954.
Fernández-Miranda C, Pérez-Carreras M, Colina F, López-Alonso
G, Vargas C, Solís-Herruzo JA. A pilot trial of fenofibrate for
the treatment of non-alcoholic fatty liver disease. Dig Liver
Dis. 2008;40:200–5.
Frades I, Andreasson E, Mato JM, Alexandersson E, Matthiesen
R, Martínez-Chantar ML. Integrative genomic signatures of
hepatocellular carcinoma derived from nonalcoholic fatty
liver disease. PLoS One. 2015;10:e0124544.
Francque S, Verrijken A, Caron S, Prawitt J, Paumelle R, Derudas
B, et al. PPAR-α gene expression correlates with severity and
histological treatment response in patients with non-alcoholic
steatohepatitis. J Hepatol. 2015;63:164–73.
Franko A, Neschen S, Rozman J, Rathkolb B, Aichler M,
Feuchtinger A, et al. Bezafibrate ameliorates diabetes via
reduced steatosis and improved hepatic insulin sensitivity in
diabetic TallyHo mice. Mol Metab. 2017;6:256–66.
Friedman SL, Neuschwander-Tetri BA, Rinella M, Sanyal AJ.
Mechanisms of NAFLD development and therapeutic strat￾egies. Nat Med. 2018;24:908–22.
Garbacz WG, Lu P, Miller TM, Poloyac SM, Eyre NS, Mayrhofer
G, et al. Hepatic overexpression of CD36 improves glycogen
homeostasis and attenuates high-fat diet-induced hepatic
steatosis and insulin resistance. Mol Cell Biol. 2016;36:
Gross B, Pawlak M, Lefebvre P, Staels B. PPARs in obesity￾induced T2DM, dyslipidaemia and NAFLD. Nat Rev
Endocrinol. 2017;13:36–49.
Haas JT, Vonghia L, Mogilenko DA, Verrijken A, Molendi-Coste
O, Fleury S, et al. Transcriptional network analysis implicates
altered hepatic immune function in NASH development and
resolution. Nat Metab. 2019;1:604–14.
Han J, Hajjar DP, Tauras JM, Feng J, Gotto AM, Nicholson AC.
Transforming growth factor-β1 (TGF-β1) and TGF-β2 de￾crease expression of CD36, the type B scavenger receptor,
through mitogen-activated protein kinase phosphorylation of
peroxisome proliferator-activated receptor-γ. J Biol Chem.
Heikkinen S, Auwerx J, Argmann CA. PPARγ in human and
mouse physiology. Biochim Biophys Acta—Mol Cell Biol
Lipids. 2007;1771:999–1013.
Holden PR, Tugwood JD. Peroxisome proliferator-activated re￾ceptor alpha: role in rodent liver cancer and species differ￾ences. J Mol Endocrinol. 1999;22:1–8.
Honda Y, Kessoku T, Ogawa Y, Tomeno W, Imajo K, Fujita K,
et al. Pemafibrate, a novel selective peroxisome proliferator￾activated receptor alpha modulator, improves the pathogene￾sis in a rodent model of nonalcoholic steatohepatitis. Sci Rep.
Cell Biol Toxicol
Ishibashi S, Arai H, Yokote K, Araki E, Suganami H, Yamashita S.
Efficacy and safety of pemafibrate (K-877), a selective per￾oxisome proliferator-activated receptor α modulator, in pa￾tients with dyslipidemia: results from a 24-week, random￾ized, double blind, active-controlled, phase 3 trial. J Clin
Lipidol. 2018;12:173–84.
Jain MR, Giri SR, Bhoi B, Trivedi C, Rath A, Rathod R, et al. Dual
PPARα/γ agonist saroglitazar improves liver histopathology
and biochemistry in experimental NASH models. Liver Int.
Kanmani P, Kim H. Protective effects of lactic acid bacteria against
TLR4 induced inflammatory response in hepatoma HepG2
cells through modulation of toll-like receptor negative regu￾lators of mitogen-activated protein kinase and NF-κB signal￾ing. Front Immunol. 2018;9:1537.
Kleiner DE, Brunt EM, Van Natta M, Behling C, Contos MJ,
Cummings OW, et al. Design and validation of a histological
scoring system for nonalcoholic fatty liver disease.
Hepatology. 2005;41:1313–21.
Kücükoglu Ö, Labenz C, Sydor S, Schlattjan M, Best J, Gerken G,
et al. Free fatty acids enhance CD36 knockdown in primary
human hepatocytes and abrogate PTEN expression—
recapitulation of NASH-associated HCC. J Hepatol.
LeBleu VS, O’Connell JT, Gonzalez Herrera KN, Wikman H,
Pantel K, Haigis MC, et al. PGC-1α mediates mitochondrial
biogenesis and oxidative phosphorylation to promote metas￾tasis. Nat Cell Biol. 2014;16:992–1003.
López-Riera M, Conde I, Tolosa L, Zaragoza Á, Castell JV,
Gómez-Lechón MJ, et al. New microRNA biomarkers for
drug-induced steatosis and their potential to predict the con￾tribution of drugs to non-alcoholic fatty liver disease. Front
Pharmacol. 2017;8:1–12.
Mandard S, Müller M, Kersten S. Peroxisome proliferator￾activated receptor α target genes. Cell Mol Life Sci.
Mannaerts I, Leite SB, Verhulst S, Claerhout S, Eysackers N,
Thoen LFR, et al. The hippo pathway effector YAP controls
mouse hepatic stellate cell activation. J Hepatol. 2015;63:
Marchesini G, Day CP, Dufour JF, Canbay A, Nobili V, Ratziu V,
et al. EASL-EASD-EASO clinical practice guidelines for the
management of non-alcoholic fatty liver disease. J Hepatol.
Maréchal L, Laviolette M, Rodrigue-Way A, Sow B, Brochu M,
Caron V, et al. The CD36-PPARγ pathway in metabolic
disorders. Int J Mol Sci. 2018;19:1–16.
Marra F, Tacke F. Roles for chemokines in liver disease.
Gastroenterology. 2014;147:577–94.
Miquilena-Colina ME, Lima-Cabello E, Sánchez-Campos S,
García-Mediavilla MV, Fernández-Bermejo M, Lozano￾Rodríguez T, et al. Hepatic fatty acid translocase CD36
upregulation is associated with insulin resistance,
hyperinsulinaemia and increased steatosis in non-alcoholic
steatohepatitis and chronic hepatitis C. Gut. 2011;60:1394–
Moylan CA, Pang H, Dellinger A, Suzuki A, Garrett ME, Guy
CD, et al. Hepatic gene expression profiles differentiate pre￾symptomatic patients with mild versus severe nonalcoholic
fatty liver disease. Hepatology. 2014;59:471–82.
Ohashi T, Tanabe J, Ishikawa T, Okumura A, Sato K, Ayada M,
et al. Inflammatory cytokines modulate chemokine produc￾tion patterns of HepG2 cells toward initially inclined direc￾tion. Hepatol Res. 2009;39:510–9.
Ratziu V, Giral P, Jacqueminet S, Charlotte F, Hartemann-Heurtier
A, Serfaty L, et al. Rosiglitazone for nonalcoholic
steatohepatitis: one-year results of the randomized placebo￾controlled fatty liver improvement with rosiglitazone therapy
(FLIRT) trial. Gastroenterology. 2008;135:100–10.
Ratziu V, Charlotte F, Bernhardt C, Giral P, Halbron M, Lenaour
G, et al. Long-term efficacy of rosiglitazone in nonalcoholic
steatohepatitis: results of the fatty liver improvement by
rosiglitazone therapy (FLIRT 2) extension trial. Hepatology.
Ratziu V, Harrison SA, Francque S, Bedossa P, Lehert P, Serfaty L,
et al. Elafibranor, an agonist of the peroxisome proliferator￾activated receptor-α and -δ, induces resolution of nonalco￾holic steatohepatitis without fibrosis worsening.
Gastroenterology. 2016;150:1147–59.
Rodrigues RM, De Kock J, Branson S, Vinken M, Meganathan K,
Chaudhari U, et al. Human skin-derived stem cells as a novel
cell source for in vitro hepatotoxicity screening of pharma￾ceuticals. Stem Cells Dev. 2014;23:44–55.
Rodrigues RM, Branson S, De Boe V, Sachinidis A, Rogiers V, De
Kock J, et al. In vitro assessment of drug-induced liver
steatosis based on human dermal stem cell-derived hepatic
cells. Arch Toxicol. 2016a;90:677–89.
Rodrigues RM, Heymans A, De Boe V, Sachinidis A, Chaudhari
U, Govaere O, et al. Toxicogenomics-based prediction of
acetaminophen-induced liver injury using human hepatic cell
systems. Toxicol Lett. 2016b;240:50–9.
Rogue A, Lambert C, Jossé R, Antherieu S, Spire C, Claude N,
et al. Comparative gene expression profiles induced by
PPARγ and PPARα/γ agonists in human hepatocytes.
PLoS One. 2011;6:e18816.
Rogue A, Anthérieu S, Vluggens A, Umbdenstock T, Claude N,
De la Moureyre-Spire C, et al. PPAR agonists reduce
steatosis in oleic acid-overloaded HepaRG cells. Toxicol
Appl Pharmacol. 2014;276:73–81.
Sanyal AJ, Chalasani N, Kowdley KV, McCullough A, Diehl AM,
Bass NM, et al. Pioglitazone, vitamin E, or placebo for
nonalcoholic steatohepatitis. N Engl J Med. 2010;362:
Sato O, Kuriki C, Fukui Y, Motojima K. Dual promoter structure
of mouse and human fatty acid translocase/CD36 genes and
unique transcriptional activation by peroxisome proliferator￾activated receptor α and γ ligands. J Biol Chem. 2002;277:
Scarpulla RC. Metabolic control of mitochondrial biogenesis
through the PGC-1 family regulatory network. Biochim
Biophys Acta. 2011;1813:1269–78.
Sumida Y, Yoneda M. Current and future pharmacological thera￾pies for NAFLD/NASH. J Gastroenterol. 2018;53:362–76.
Tacke F, Zimmermann HW, Trautwein C, Schnabl B. CXCL5
plasma levels are decreased in patients with chronic liver
disease. J Gastroenterol Hepatol. 2011;26:523–9.
Tarek I, Tamini A-R, Elgouhari HM, Alkhouri N, Yerian LM,
Berk MP, et al. An apoptosis panel for nonalcoholic
steatohepatitis diagnosis. J Hepatol. 2011;54:1224–9.
Ten RM, Paya CV, Israel N, Le Bail O, Mattei MG, Virelizier JL,
et al. The characterization of the promoter of the gene
Cell Biol Toxicol
encoding the p50 subunit of NF-κB indicates that it partici￾pates in its own regulation. EMBO J. 1992;11:195–203.
Thomas E, Gonzalez VD, Li Q, Modi AA, Chen W, Noureddin M,
et al. HCV infection induces a unique hepatic innate immune
response associated with robust production of type III inter￾ferons. Gastroenterology. 2012;142:978–88.
Tong L, Wang L, Yao S, Jin L, Yang J, Zhang Y, et al. PPAR δ
attenuates hepatic steatosis through autophagy-mediated fatty
acid oxidation. Cell Death Dis. 2019;10:1–14.
Vaidyanathan S, Maboudian M, Warren V, Yeh C, Dieterich HA,
Howard D, et al. A study of the pharmacokinetic interactions
of the direct renin inhibitor aliskiren with metformin, pioglit￾azone and fenofibrate in healthy subjects. Curr Med Res
Opin. 2008;24:2313–26.
Varga ZV, Ferdinandy P, Liaudet L, Pacher P. Drug-induced mi￾tochondrial dysfunction and cardiotoxicity. Am J Physiol—
Heart Circ Physiol. 2015;309:1453–67.
Wettstein G, Luccarini J-M, Poekes L, Faye P, Kupkowski F,
Adarbes V, et al. The new-generation pan-peroxisome
proliferator-activated receptor agonist IVA337 protects the
liver from metabolic disorders and fibrosis. Hepatol
Commun. 2017;1:524–37.
Xu L, Hui AY, Albanis E, Arthur MJ, Blaner WS, Mukherjee P,
et al. Human hepatic stellate cell lines, LX-1 and LX-2: new
tools for analysis of hepatic fibrosis. Gut. 2005;54:142–51.
Yan F, Wang Q, Xu C, Cao M, Zhou X, Wang T, et al. Peroxisome
proliferator-activated receptor α activation induces hepatic
steatosis, suggesting an adverse effect. PLoS One. 2014;9:
Yang L, Roh YS, Song J, Zhang B, Liu C, Loomba R, et al. TGF-β
signaling in hepatocytes participates in steatohepatitis
through regulation of cell death and lipid metabolism.
Hepatology. 2014;59:483–95.
Younossi Z, Anstee QM, Marietti M, Hardy T, Henry L, Eslam M,
et al. Global burden of NAFLD and NASH: trends, predic￾tions, risk factors and prevention. Nat Rev Gastroenterol
Hepatol. 2018;15:11–20.
Publisher’s note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional
Cell Biol Toxicol

Leave a Reply

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>