an interspecies comparison of toxicity pathways mediating

AN INTERSPECIES COMPARISON OF TOXICITY PATHWAYS MEDIATING NEURODEVELOPMENTAL TOXICITY IN NEUROSPHERES - AND NOVEL COMPUTATIONAL APPROACHES FOR HIGH CONTENT IMAGE ANALYSES (HCA)

Ellen Fritsche EPA’s Computational Toxicology Communities of Practice September 25th 2014

Cell Biological Processes performed by NPC

With courtesy from William Mundy, U.S. Environmental Protection Agency and John Havel, SRA International, Inc.

Epi-genetics

The Neurosphere Method

Fetal Brain Culture as Neurospheres

NPC Proliferation: -Increase in size/time -BrdU incorporation

NPC Differentiation: -Neurons -Astrocytes -Oligodendrocytes

NPC Migration: -Total migration distance -number of migrated cells -neuronal migration

Neural Stem/Progenitor Cells (NPCs, Lonza)

Fetal Brain

Neural Stem/Progenitor Cells self-prepared

From Breier…Fritsche.. et al. Neurotoxicol Teratol 2009

The ‘Neurosphere-‐Assay’

Apoptosis

BrdU

Fritsche et al. Environ Health Perspect 2005

Moors et al. Toxicol Appl Pharmacol 2007

Moors et al. Environ Health Perspect 2009

Moors et al. Genes & Immunity 2010

Tegenge et al. Cell. Mol. Life Sci. 2010

Schreiber et al. Environ Health Perspect 2010

Gassmann et al. Environ Health Perspect 2010

Verner et al. Toxicol in Vitro 2011

Fritsche et al. Methods Mol Biol 2011

Gassmann et al. Toxicol in Vitro 2012

Bal-Price et al. ALTEX 2012

Baumann et al. Curr. Protoc. Toxicol. 2014

Gassmann et al. Arch. Toxicol. 2014

Fritsche Methods Pharmacol. Toxicol. 2014

Alépée et al. ALTEX 2014

3 days

5 days

Outline

•  Thyroid Hormone (TH)

•  Arylhydrocarbon Receptor (AhR)

•  Valproic Acid

-  Histone Deacetylase (HDAC) inhibition

-  Apoptosis

•  HCA (Cellomics Arrayscan & self-written

algorithms ‘Omnisphero’)

Thyroid Hormone

E11-‐E14 =

GW 6-‐9 P1-‐7 =

GW 18-‐30 P8-‐14 =

GW 32-‐40

Perez-Castillo et al, Endocrinology 117:2457-2461,1985

Ontogeny of T3 receptors in rat

T H R a1 e x p re s s io nC

op

y n

um

be

rs o

f T

HRa

1/

10

00

0 ß

Ac

tin

P r o l 2 4 h D iff0

5 0

1 0 04 0 0

6 0 0

8 0 0

1 0 0 0h N P C (n = 3 )

m N P C (n = 3 )

α1 α

1 T H R b1 e x p re s s io n

Co

py

nu

mb

ers

of

TH

Rb

1/

10

00

0 ß

Ac

tin

P r o l 2 4 h D iff0

5

1 0

1 5

2 0h N P C (n = 3 )

m N P C (n = 3 )

β

β

www.translatingtime.net

Thyroid Hormone

O4

+ c

ell

s [

% c

on

tro

l]

C T 3 T 4 T 3 T 40

5 0

1 0 0

1 5 0

2 0 0

*

O lig o d e n d ro g e n e s isH a ir le s s e x p re s s io n

Co

py

nu

mb

ers

of

Ha

irle

ss

/10

00

0 ß

Ac

tin

C 3 n M T 30

5 0

1 0 0

1 5 0

2 0 0

2 5 0h N P C

m N P C

*

*

O lig o d e n d ro g e n e s is o f T H R b- /- m N P C s

O4

+ c

ell

s [

% c

on

tro

l]

0

5 0

1 0 0

1 5 0

2 0 0W ild ty p e (n = 4 )

T H R b- /- (n = 1 )

1 0 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

*

*

β

β

O lig o d e n d ro g e n e s is o f T H R a- /- m N P C s

O4

+ c

ell

s [

% c

on

tro

l]

0

5 0

1 0 0

1 5 0

2 0 0 W ild ty p e (n = 4 )

T H R a - /- (n = 2 )

1 0 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

*

*

α

α

H u m a n o lig o d e n d ro g e n e s is

O4

+ c

ell

s [

% c

on

tro

l]

0

5 0

1 0 0

1 5 0

2 0 0

2 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

* *

O4 Hoechst

Transgenic animals kindly provided by Heike Heuer, IUF

Thyroid Hormone

BDE-99

T3 OL formation

OL maturation

THRα

BDE-99

M O G e x p re s s io n - m N P C

DD

CT

[no

rme

d t

o C

of

da

y 1

]

0

5

1 0

P ro life ra tin g

1 d d iffe re n tia t in g



1 0 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

M B P e x p re s s io n - h N P C

DD

CT

[no

rme

d t

o C

of

da

y 1

]

02468

1 0

2 02 53 03 54 0 P ro life ra tin g




2 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

*

#

IC50 = 14 µM

BDE-99

T3 OL formation

OL maturation

THRα

BDE-99

X

M O G e x p re s s io n - m N P C

DD

CT

[no

rme

d t

o C

of

da

y 1

]

0

5

1 0

P ro life ra tin g




1 0 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

M B P e x p re s s io n - h N P C

DD

CT

[no

rme

d t

o C

of

da

y 1

]

02468

1 0

2 02 53 03 54 0 P ro life ra tin g




2 µM B D E -9 9

3 n M T 3

-

-

+

-

-

+

+

+

*

#

Thyroid Hormone

IC50 = 2 µM


1 10 10 0.1 1 0.750

25

50

75

100

125

*

µM3-MC Bap MNF MeHgClDMSO

% m

igra

tio

n

AhR Agonists

AhR Antagonist AhR

Agonists AhR

Antagonist

AhR -‐ migraOon

Gassmann et al.

1574 VOLUME 118 | NUMBER 11 | November 2010 Environmental Health Perspectives

by 3-MC or B(a)P (Figure 2D). Interestingly, TCDD did not disrupt wild-type mNPC migration despite AhR activation (Figure 2D). This might be due to the fact that, in con-trast to 3-MC and B(a)P, TCDD is hardly metabolized and thus does not produce reactive intermediates.

AhR-dependent gene transcription. AhR-dependent gene transcription is inducible only in mNPCs, and not in hNPCs because of low abundance of AhR and ARNT tran-scripts and absence of AhR protein in human cells. Because 3-MC, B(a)P, and MNF did not influence hNPC viability, proliferation,

or migration but did modulate proliferation or migration of mNPCs, we determined AhR and ARNT mRNA expression after exposure to these compounds in hNPCs and mNPCs under proliferating and di!erentiating condi-tions. AhR and ARNT mRNAs were expressed at very low copy numbers in hNPCs (0.6–2.5 and 13/1,000 copies -actin, respectively) and higher copy numbers in mNPCs (20–63 and 716–1,045/1,000 copies -actin, respec-tively) (Table 1). In addition, CYP1A1 expres-sion was undetectable in untreated hNPCs. That -actin was in this case valid to use as a house keeping gene was demonstrated by

normalization of proliferating versus differ-entiating NPCs to three additional house-keeping genes [RPL27, RPL30, OAZ1; see Supplemental Material, Figure 2 (doi:10.1289/ehp.0901545)]. Expression of the AhR target genes AhRR, CYP1A1, CYP1B1, and c-myc was not significantly induced by 10 µM 3-MC after 6, 12, 24, and 48 hr of di!erentiation in hNPCs (Figure 3A). We obtained comparable results for hNPCs from a second individual after 6 hr of treatment (data not shown). In contrast, 6 hr of exposure to 10 µM 3-MC significantly induced Cyp1a1 and Cyp1b1 mRNA in wild-type mNPCs to levels 6.6 ± 1.7 and 2.5 ± 0.25 times higher than con-trols, respectively (Figure 3C). Although 1 nM TCDD did not disturb neural migration, it induced AhR signaling in wild-type mNPCs, increasing Cyp1a1 expression 21 times relative to controls (Figure 3C, inset).

Comparison of mRNA expression lev-els between hNPCs and mNPCs showed that genes belonging to the AhR machin-ery and AhR-dependent genes were gener-ally expressed in higher copy numbers/1,000 copies -actin in mNPCs than in hNPCs (Table 1). Lack of AhR protein in hNPCs was confirmed by Western blot (Figure 3B). "ese results demonstrate that AhR signaling pathway gene products mediate the e!ects of the AhR agonists 3-MC and B(a)P and the AhR antagonist MNF on proliferation and migration in mNPCs, because AhR-deficient mNPCs are protected against these effects (Figure 3C, inset).

Discussion"e development of cell-based, non animal test-ing strategies for hazard assessment of chemicals is currently one of the most important tasks in toxicological research. In this regard, it is most important to choose appropriate model systems that are truly predictive for humans (Krewski et al. 2009; National Research Council 2007). Human tumor cell lines that are easily acces-sible in large quantities bear the restriction that they do not represent cellular metabolism and signal transduction of normal cells. In con-trast, primary cells are often obtained as ex vivo cultures from rodents. Such primary cultures are regarded as superior to tumor-derived cells. However, species-specific differences limit their application. One example of how rodent primary cells can indeed mis classify hazards for humans is provided by this study, which shows mouse-derived primary cells to be more susceptible to AhR modulation than their human counter parts. With regard to chemical testing, it is critical to be aware of such di!erences in order to avoid over- or under estimating hazards that chemicals pose to humans, and thereby protect human health and allow industry production and develop-ment of chemicals at the same time.

Figure 2. AhR agonists shorten wild-type mNPC but not hNPC migration. hNPCs (A and C) and wild-type (B and D) and AhR-KO (D) mNPCs were exposed to 3-MC, B(a)P, TCDD, MNF, or MeHgCl (PC) during differen-tiation for 48 hr. Migration distance from the edge of the sphere to the furthest outgrowth was measured. Data represent means ± SEs of two (AhR-KO) to five independent experiments (5–8 spheres/exposure). Bar = 500 μm. *p < 0.05 compared with 0.1% DMSO.

125

100

75

50

25

0

125

100

75

50

25

0

Mig

ratio

n (%

)

Mig

ratio

n (%

)

*

*

**

*

13-MC TCDD

DMSO

WT AhR-KO

Treatment (µM) Treatment (µM)

MNF MeHgCI MeHgCIDMSO

3-MC MNF10 10 10 100.001 0.0010.1 0.11 1 1 11

Human

Human

Mouse

Mouse

B(a)PTCDDB(a)P

Table 1. Comparison of human and mouse mRNA copy numbers/1,000 copies of -actin in proliferating and 24-hr differentiating NPCs.a

Human Mouse Mouse:human ratioGene Prolif Diff Prolif Diff Prolif DiffAhR 2.45 0.62 20.08 63.21 8.19 102.12ARNT 12.95 13.55 716.26 1045.07 55.31 77.13AhRR 0.94 0.61 79.21 387.55 84.24 638.85CYP1A1 < 0.001 < 0.001 18.63 31.64 ND NDCYP1B1 0.01 0.06 146.24 834.91 16407.23 14990.75C-MYC 5559.59 3017.13 14835.55 15967.72 2.67 5.29Abbreviations: Diff, differentiating; ND, not detectable; Prolif, proliferating.aData represent at least three independent experiments. The mouse:human ratio is shown to compare human and mouse mRNA expression levels.

Gassmann et al.

1574 VOLUME 118 | NUMBER 11 | November 2010 Environmental Health Perspectives

by 3-MC or B(a)P (Figure 2D). Interestingly, TCDD did not disrupt wild-type mNPC migration despite AhR activation (Figure 2D). This might be due to the fact that, in con-trast to 3-MC and B(a)P, TCDD is hardly metabolized and thus does not produce reactive intermediates.

AhR-dependent gene transcription. AhR-dependent gene transcription is inducible only in mNPCs, and not in hNPCs because of low abundance of AhR and ARNT tran-scripts and absence of AhR protein in human cells. Because 3-MC, B(a)P, and MNF did not influence hNPC viability, proliferation,

or migration but did modulate proliferation or migration of mNPCs, we determined AhR and ARNT mRNA expression after exposure to these compounds in hNPCs and mNPCs under proliferating and di!erentiating condi-tions. AhR and ARNT mRNAs were expressed at very low copy numbers in hNPCs (0.6–2.5 and 13/1,000 copies -actin, respectively) and higher copy numbers in mNPCs (20–63 and 716–1,045/1,000 copies -actin, respec-tively) (Table 1). In addition, CYP1A1 expres-sion was undetectable in untreated hNPCs. That -actin was in this case valid to use as a house keeping gene was demonstrated by

normalization of proliferating versus differ-entiating NPCs to three additional house-keeping genes [RPL27, RPL30, OAZ1; see Supplemental Material, Figure 2 (doi:10.1289/ehp.0901545)]. Expression of the AhR target genes AhRR, CYP1A1, CYP1B1, and c-myc was not significantly induced by 10 µM 3-MC after 6, 12, 24, and 48 hr of di!erentiation in hNPCs (Figure 3A). We obtained comparable results for hNPCs from a second individual after 6 hr of treatment (data not shown). In contrast, 6 hr of exposure to 10 µM 3-MC significantly induced Cyp1a1 and Cyp1b1 mRNA in wild-type mNPCs to levels 6.6 ± 1.7 and 2.5 ± 0.25 times higher than con-trols, respectively (Figure 3C). Although 1 nM TCDD did not disturb neural migration, it induced AhR signaling in wild-type mNPCs, increasing Cyp1a1 expression 21 times relative to controls (Figure 3C, inset).

Comparison of mRNA expression lev-els between hNPCs and mNPCs showed that genes belonging to the AhR machin-ery and AhR-dependent genes were gener-ally expressed in higher copy numbers/1,000 copies -actin in mNPCs than in hNPCs (Table 1). Lack of AhR protein in hNPCs was confirmed by Western blot (Figure 3B). "ese results demonstrate that AhR signaling pathway gene products mediate the e!ects of the AhR agonists 3-MC and B(a)P and the AhR antagonist MNF on proliferation and migration in mNPCs, because AhR-deficient mNPCs are protected against these effects (Figure 3C, inset).

Discussion"e development of cell-based, non animal test-ing strategies for hazard assessment of chemicals is currently one of the most important tasks in toxicological research. In this regard, it is most important to choose appropriate model systems that are truly predictive for humans (Krewski et al. 2009; National Research Council 2007). Human tumor cell lines that are easily acces-sible in large quantities bear the restriction that they do not represent cellular metabolism and signal transduction of normal cells. In con-trast, primary cells are often obtained as ex vivo cultures from rodents. Such primary cultures are regarded as superior to tumor-derived cells. However, species-specific differences limit their application. One example of how rodent primary cells can indeed mis classify hazards for humans is provided by this study, which shows mouse-derived primary cells to be more susceptible to AhR modulation than their human counter parts. With regard to chemical testing, it is critical to be aware of such di!erences in order to avoid over- or under estimating hazards that chemicals pose to humans, and thereby protect human health and allow industry production and develop-ment of chemicals at the same time.

Figure 2. AhR agonists shorten wild-type mNPC but not hNPC migration. hNPCs (A and C) and wild-type (B and D) and AhR-KO (D) mNPCs were exposed to 3-MC, B(a)P, TCDD, MNF, or MeHgCl (PC) during differen-tiation for 48 hr. Migration distance from the edge of the sphere to the furthest outgrowth was measured. Data represent means ± SEs of two (AhR-KO) to five independent experiments (5–8 spheres/exposure). Bar = 500 μm. *p < 0.05 compared with 0.1% DMSO.

125

100

75

50

25

0

125

100

75

50

25

0

Mig

ratio

n (%

)

Mig

ratio

n (%

)

*

*

**

*

13-MC TCDD

DMSO

WT AhR-KO

Treatment (µM) Treatment (µM)

MNF MeHgCI MeHgCIDMSO

3-MC MNF10 10 10 100.001 0.0010.1 0.11 1 1 11

Human

Human

Mouse

Mou

se

B(a)PTCDDB(a)P

Table 1. Comparison of human and mouse mRNA copy numbers/1,000 copies of -actin in proliferating and 24-hr differentiating NPCs.a

Human Mouse Mouse:human ratioGene Prolif Diff Prolif Diff Prolif DiffAhR 2.45 0.62 20.08 63.21 8.19 102.12ARNT 12.95 13.55 716.26 1045.07 55.31 77.13AhRR 0.94 0.61 79.21 387.55 84.24 638.85CYP1A1 < 0.001 < 0.001 18.63 31.64 ND NDCYP1B1 0.01 0.06 146.24 834.91 16407.23 14990.75C-MYC 5559.59 3017.13 14835.55 15967.72 2.67 5.29Abbreviations: Diff, differentiating; ND, not detectable; Prolif, proliferating.aData represent at least three independent experiments. The mouse:human ratio is shown to compare human and mouse mRNA expression levels.

1"

10"

100"

1000"

10000"

100000"

1000000"

AhR" ARNT" CYP1A1" CYP1B1"

*

*

mRNA


7 x

3 x

What is the reason for species-‐specific differences?


hNPC

/105

co

pie

s β-

Act

in

3-‐MC DMSO

human mouse

HepG2

GAPDH

AhR

Protein

AhR -‐ gene/protein expression

Human NPC are protected against AhR-dependent toxicity of PAH due to lack of AhR expression

?

Valproic Acid

↑ ↑

↓ Molecular mechanisms: Prolifera>on: wnt – β-‐catenin (Fo> et al. 2013) Apoptosis: Bcl-‐2 expression (Go et al. 2011) Differen>a>on: suggested indirectly by increasing the NPC pool? HDAC inhibi>on suggested (Montgomery et al. 2004)

? ↑

Apoptosis: Bcl-‐2 expression (Go et al. 2011)

Apoptosis

Apoptosis

E11-‐E14 =

GW 6-‐9 P1-‐7 =

GW 18-‐30 P8-‐14 =

GW 32-‐40

↓

↓ Fo> et al. 2013

↑ Apoptosis

Yochum et al. 2011

Mech.: HDAC inhibi>on

Valproic Acid -‐ NPC proliferaOon

NPC proliferation

RAT

HUMAN

Baumann et al. in preparation

IC50 = 380 µM

IC50 = 756 µM

Valproic Acid-‐ NPC differenOaOon

NPC differentiation RAT

HUMAN

Baumann et al. in preparation

IC50 = 321 µM

IC50 = 3177 µM

↓

Valproic Acid

↑ ↑

↓ ↓ ↑ Apoptosis

Apoptosis

E11-‐E14 =

GW 6-‐9 P1-‐7 =

GW 18-‐30 P8-‐14 =

GW 32-‐40

↓ ↓

↑ Apoptosis

•  NPC prolifera>on is inhibited by VPA-‐dependent HDAC inhibi>on (Baumann et al. in prep)

•  VPA does not inhibit neuronal differen>a>on of rat NPC, but induces neuronal apoptosis due to forma>on of ROS.

•  Human neurons are protected against VPA-‐induced apoptosis.

Long-Term Culture (25d) of hNPCs in Mal-PVA Hydrogels (Cellendis) (Hellwig et al. in preparation)

200 x mag. 200 x mag. Phalloidin, Hoechst, ß-III tubulin

Summary (I)

•  Primary Neurospheres seem to conserve NPC molecular signatures and functions ex vivo into in vitro.

•  Due to the multiple neurodevelopmental processes neurospheres are apt to mimic in vitro, they are well suited for investigations ‘from pathway to function’.

•  A variety of interspecies differences in NPC signaling seem to exist between human and rodent NPCs.

•  Studying the molecular similarities/differences of neurospheres will contribute to human risk assessment for DNT, especially in the context of the ‘Adverse Outcome Pathway’ concept.

HCA in differenOaOng NPC

Ø  Information on the whole migration area, not just random areas of a well.

Ø  Software needs to distinguish between different cell types.

Ø  Issue of a high density culture needs to be overcome and neurons correctly identified.

Ø  Varying densities in different areas around sphere core.

196x

From: Breier et al., Toxicological Science, 2008 Harrill et al., Neuro Toxicology, 2010 Harrill et al., Toxicology in vitro, 2011

Synaptogenesis

Proliferation/ Viability

Neurite Outgrowth

High Content Image Analyses (HCA) for DNT tesOng

HCA in differenOaOng NPC

Neuronal Quantification

Neuronal Morphology

Migration

Neuronal Positioning

Neuron/Nuclei Identification

Schmuck et al. in preparation

Thresholding (Isodata) Watershade

Thresholding (Isodata)

Raw images Binary images

Omnisphero: Image Pre-‐processing


Optional: Manual annotation Composite Fill Neuron Tracer


Omnisphero: Neuron IdenOficaOon

Omnisphero: Automated Analyses


EC50-Values EGF [ng/ml]

Acrylamide [mM]

MeHgCl [µM]

Manual 0,96 0,30 0,045

Neuronal Profiling 1,442 0,34 0,049

Composite Fill 0,67 0,31 0,104

Neuron Tracer 1,022 0,41 0,051

Neurogenesis: comparison of methods (IC50 values)

Acrylamide EGF Methylmercury


Neurogenesis: Accuracy & Precision

Acrylamide EGF Methylmercury

Detection Power (DP)

False-Positives (FP)


Neurite outgrowth: Accuracy & Precision


Floodfill-Filter

Program written by Thomas Temme

Sphere-‐specific Endpoints

Sphere-‐specific Endpoints: MigraOon distance

∑= n

n

n

n

n

n

Nuclein

NeuronsnNuclein

Neuronsn

1 )()(

)()(

σ

R in g n u m b e r

Ne

uro

n D

en

sit

y

1 2 3 4 5 6 7 8 90 .0

0 .5

1 .0

1 .5

2 .0

Neuronal Density DistribuOon

Neuronal Density DistribuOon

v

Summary (II)

•  The Neurosphere assay needs specific algorithms for HCA: development of ‘Omnisphero’.

•  Composite fill and Neuron tracer significantly improve true neuronal detection rate as a ‘conventional’ HCA endpoint.

•  Sphere-specific endpoints like migration and neuronal positioning can be assessed by ‘Ominsphero’.

Dr. Marta Barenys Dr. Julia Tigges Dr. Susanne Giersiefer Dr. JaneZe Schuwald Dr. Henrik Alm Jenny Baumann Katharina Dach Mar>n Schmuck Maxi Hofrichter Stefan Masjosthusmann Chris>ne Hellwig Chris>ane Hohensee Laura Nimtz Denise de Boer Ulrike Hübenthal

Thank you for your aaenOon!

CooperaOon partners: Pamela Lein, UC Davis, USA Heike Heuer, IUF, Düsseldorf Axel Mosig, University of Bochum Kai Stühler, HHU Düsseldorf

Acknowledgements

LRI Innovative Science Award 2006 German Alternative

Methods Award 2007

Leibniz - DAAD�

AOP – DNT:

DNT – TranslaOng Time

www.translatingtime.net Processes: Brain Growth & Neurogenesis, whole brain

GW HUMAN

PC

DA

YS

1st trimester 3rd trimester 2nd trimester

0

5

10

15

20

25

30

35

4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44

RAT

MOUSE

birth

birth

Clancy et al. Neuroinformatics 2007 Nagarajan et al. Neuroinformatics 2010

an interspecies comparison of toxicity pathways mediating

Documents