©2013 For research use only. Not for use in diagnostic procedures. Trademarks mentioned herein are property of Molecular Devices, LLC or their respective owners.

Predictive High-Throughput Assays for Hepatotoxicity and Cardiac

Physiology Using iPSC-Derived Cell ModelsOksana Sirenko1, Jayne Hesley1, Carole Crittenden1, Vanessa Ott2, Blake Anson2, Ivan Rusyn3, and Evan F. Cromwell1

1Molecular Devices, LLC, 1311 Orleans Drive, Sunnyvale, CA 94089, 2Cellular Dynamics International, 525 Science Drive, Madison WI 537113 University of North Carolina, Chapel Hill, NC 27599

Introduction

Materials & Methods

Summary• We developed and validated assays for measuring impact of pharmacological

compounds on beating rate of stem cell derived cardiomyocytes with FLIPR Tetra

System or SpectraMax i3 and multi-parametric analysis of beating profiles. The

estimated predictive value of the assay was greater than 80%, while the multi-

parametric analysis has demonstrated prediction long QT syndrome and other

types of cardiotoxicity.

• Live-cell assays using the ImageXpress Micro XL High Content Imaging System with

human iCell Hepatocytes can measure the impact of pharmacological compounds

on hepatocyte viability and intrinsic hepatocyte functions.

• We demonstrate utility of these in vitro assay models for toxicity screening and

understanding potential hepatotoxic effects early in the drug development process.

Hepatotoxicity Library ScreeningA multi-parametric imaging based assay was used to monitor changes in cell viability (Calcein

AM), nuclear shape (Hoechst), and mitochondria integrity (MitoTracker Orange) associated

with different types of toxicity. Screen-Well™ Hepatotoxicity Library (ENZO) contains 240

compounds including anti-cancer , anti-inflammatory, neuroleptic, antibiotics, and other

classes. Compounds represent different mechanisms of hepatotoxicity: ALT elevation,

steatosis, phospholipidosis, mitochondria damage, etc. The assay was used to asses toxicity of

compounds in the library (72 hr, 5 concentrations). In addition, mitochondria

potential/oxidative stress assay (JC-10) 60min was also used to increase overall predictivity.

Output parameters:– Number of Calcein pos. cells– Number of MitoTracker pos. – Total cell number – Calcein AM intensity – Calcein AM cell area – MitoTracker intensity– MitoTracker cell area– Nuclear area– Nuclear average intensity– JC-10 Granules/Cell– JC-10 Total Granules

Toxicity Cell CountOther

Parameters

Sensitivity 40% 60%

Specificity 97% 91%

Predictive value 69% 75%

• High predictivity was found for many classes of compounds, e.g. neuroleptic, anti-

cancer, cardiac drugs, toxins.

• Lower predictivity was observed for anti-inflammatory, antibiotics, and anti-viral

drugs.

•Assay Sensitivity “hits”

based on output

parameters deviating

more than ±3σ from

DMSO controls.

•Number of compounds

in group shown in

parentheses

AC

E in

hibi

tors

(7)

Nuc

leos

ide

anal

og R

TIs

(10

)

Non

-hep

atot

oxic

(32

)

Oth

er c

lass

es (

30)

Ora

l hyp

ogly

cem

ics

(4)

Ant

i-Inf

lam

mat

ory

(23)

Ant

ibio

tics

(32)

Sta

tins

(cho

lest

erol

) (1

1)

Neu

role

ptic

s (2

7)

Hor

mon

al (

7)

Ant

i-Can

cer

(14)

Vas

cula

r (7

)

Ant

i-fun

gal (

8)

Car

diac

dru

gs (

11)

Kno

wn

toxi

ns (

13)

Ant

i-His

tam

ines

(7)

0102030405060708090

100

Ass

ay s

ensi

tivity

(%

)

We tested the Screen-Well™ Cardiotoxicity Library (ENZO Life Sciences): 107 compounds

representing different types of cardiotoxic compounds, as well as 24 non-toxic substances. We

measured the concentration-response effects of these compounds after 30 min or 24 hrs of

incubation; 4-6 concentrations were tested. Overall, 69 compounds affected beating frequency, as

well as other parameters, greater than 3 SDs from DMSO controls in at least one concentration.

Only one of the negative control compounds caused significant deviations of beating patterns at the

highest concentration. Among positive compounds were cardiac glycosides, anti-arrythmic

compounds, α- and β-adrenoreceptor blockers, Ca2+ and K+ channel blockers, dopamine receptor

antagonists, tricyclic anti-depressants, and anti-cancer drugs. The estimated predictive value of the

assay was greater than 80%, while the multi-parametric analysis has demonstrated great value in

prediction long QT syndrome and other types of arrhythmic and non-arrhythmic cardiotoxicity.

Left: Results from screening a library of cardiotoxic

(top) and “safe” (bottom) compounds. Color code

shows deviations from DMSO controls. Clustering

order set using “peak width” at 30 min.

Right: Principal component analysis using chemical

descriptors (top), or responses in the

cardiomyocyte beating assay at 30 min and 24 hrs.

Calcein AM, MitoTracker Orange, Hoechst

3µM 10µMcontrol

Cardiotoxicity Library Screening

1 2 3 4

Control Epinephrine Cisapride

1

23

4

Inte

nsity

Time

Beats per Minute

BPM + Peak Width

High Content Toxicity

Sensitivity 66% 86% 58%Specificity 100% 93% 92%Predictivity 82% 89% 75%

Cardiomyocyte Beating Assay using Intracellular Ca2+ Flux

iPSC derived iCell® Cardiomyocytes

•Cells were received frozen from Cellular Dynamics International (CDI). Cells were thawed

and plated according to CDI protocol at 20K/96well plate or 4 K/384 well on gelatin coated

plates and incubated for typically 7 days. The presence of strong synchronous contractions of

cells in the wells under the light microscope was confirmed prior to running experiments.

FLIPR® Tetra System Ca2+ Flux Assay

•The assays employ FLIPR® Calcium 5 dye (Molecular Devices), a calcium sensitive dye, to

monitor changes in Ca2+ fluxes synchronous with cell beating. Reagents were added to the

plates and incubated 30 min at 37oC in 5% CO2.

•Experimental plates were loaded and a pre-drug read was acquired. Acquisition was set up

for FLIPR Calcium 5 dye (Ex 485nm, Em 530nm). Data acquired at ~8 fps

•FLIPR Tetra System reads were acquired during compound addition and at prescribed times

after addition (read time ~ 2 min).

•Data was analyzed using FLIPR Screenworks® Peak Pro™ software to automatically detect

and characterize individual contraction events.

High Content Imaging

•Images were acquired with an ImageXpress® Micro XL System using 20x, 10x, or 4x objectives.

•The following filters were used

• Calcein AM, Cyto-ID, Neutral Lipids: FITC Filter Cube

• MitoTracker Orange, Phospholipids: TRITC Filter Cube

• Hoechst: DAPI Filter Cube.

Cell Preparation

• iPSC-derived hepatocytes (iCell® Hepatocytes) from Cellular Dynamics International (CDI)

were plated according to their recommended protocol.

• Cells were plated at a density of 60K/well (96-well plate) or 15K/well (384-well plate) on

collagen coated plates and incubated for 2-3 days. Then cells were treated with appropriate

compounds for 72 hr.

Cardiotoxicity Using SpectraMax® i3 System with MiniMax Option

Isoproterenol10 µµµµM

Control Haloperidol 10 µµµµM

Doxorubicin 30 µµµµM

Cell viability determination via imaging with live-cell stains can provide complementary information to that

obtained by the cardiac beating assay. We have developed capabilities on the new SpectraMax i3 multi-

mode plate reader to do fast kinetic fluorescence measurements of contracting cardiomyocytes. Beat rate

and peak parameters are automatically calculated by the Peak Pro algorithms contained in the SoftMax Pro

plate reader control software. In addition the SpectraMax MiniMax imaging cytometer option allows images

to be acquired of the same wells to provide phenotypic information as well as concentration response of cell

viability.

Development of highly predictive in vitro assays is extremely important for improving the

drug development process and reducing drug attrition due to toxicity. iPSC-derived cell

models are rapidly being adopted by the pharmaceutical industry for pre-clinical toxicity

studies. iPSC-derived cells provide relevant human biology and are increasingly being

shown to accurately predict drug-induced toxicity. Human iPSC-derived hepatocytes and

cardiomyocytes show great promise with respect to primary tissue-like phenotype,

unlimited availability, and a potential to establish cells from different individuals. To enable

the full potential of iPSC-derived cell models, it is necessary to develop highly predictive in

vitro assays that can be performed in a high throughput manner.

We have developed several high content imaging (HCI) assays to assess drug-induced

hepatotoxicity and cardiotoxicity. First, we used HCI and iPSC-derived hepatocytes to assess

general and mechanism-specific liver toxicity. We have also developed a physiological assay

for measuring the impact of pharmacological compounds on the beating rate of human

iPSC-derived cardiomyocytes with fast kinetic fluorescence imaging systems.

Cardiomyocyte contraction rate and pattern was characterized via changes in intracellular

Ca2+ measured using calcium sensitive dyes. These assays have been adapted for 384 well

plates and automated protocols were developed to use for small library screens.

Top: Time lapse series of images of contracting cardiomyocytes loaded with a Ca2+ sensitive dye.

Bottom: Integrated intensity versus time showing beating behavior for three different conditions.

Relative timing of images with respect to a contraction event is indicated numerically.

Images of iCell

Hepatocytes treated with

indicated concentrations

of amitryptiline for 72 hr.

Images were taken with

ImageXpress Micro XL

system using a 10x

objective.

Images and beat traces of iCell Cardiomyocytes

treated as indicated for 30 min. Data acquired

on SpectraMax i3 system with MiniMax option.

Compound Profiling

JC -10Granules Calcein AM NuclearPositive

cell countJC -10

Ln (

fold

ch

ange

vs

DM

SO

)

-16

-14

-12

-10

-8

-6

-4

-2

0

2

1 2 3 4 5 6 7 8 9

Lamivudine

Miconazole

Simvastatin

Aflatoxin B1

Leflunomide

Nimesulide

Flecainide

Amiodarone

Daunorubicin

Mitomycin C

Chlorprothixene

Prochloperazine

No/

cell

Tot

al

Tot

al A

rea

Ave

Are

a

Inte

nsity

Ave

Are

aA

ve I

nten

sity

Cal

cein

AM

Mito

Tra

cker

Tot

al c

ells

No/

cell

Tot

al

Assay Sensitivity by Compound Class