tox21 update: relevance of tox21 to nutrition - ilsi...

The Hamner Institutes for Health Sciences | ILSI FNSP Meeting | July 10, 2013 0

July 10, 2013

ILSI North America Food Nutrition & Safety Program

Russell Thomas, Ph.D.

Director, Institute for Chemical Safety Sciences

The Hamner Institutes for Health Sciences

Tox21 Update: Relevance of Tox21 to Nutrition


Broad-Based Movement in Toxicology

Towards In Vitro Testing


• A transformative redefinition of toxicity

testing is required to meet key design criteria

and take advantage of modern biology. • Broader coverage of chemicals, mixtures,

end-points, and life-stages

• Reduce cost and time

• Use fewer animals and reduce suffering

• More robust basis for risk assessment by

applying mechanistic data and relevant doses

• … a not-so-distant future where all routine

toxicity testing will be conducted in human

cells or cell lines in vitro by evaluating

perturbations of cellular responses in a suite

of toxicity pathway assays.

One Vision for Transforming Toxicity Testing


1

10

100

1000

10000

Nu

mb

er

of

Ch

em

icals

IRIS TRI

Pesticides Inerts

CCL 1 & 2 HPV

MPV

0

10

20

30

40

50

60

70

Pe

rce

nt

of

Ch

em

icals

Acute Cancer

Gentox Dev Tox

Repro Tox Judson, et al EHP (2010)

1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 | | | | | | | | |

Acute toxicity studies

(LD50) developed to

standardized batches

of pharmaceuticals

Genotoxicity

assays

developed

Draize test

introduced for

eye irritants

Rodent cancer

bioassay

introduced

Continuous

breeding studies

for reproductive

toxicity

Reasons Underlying Effort to Modernize Tox

Testing

... and cannot efficiently assess safety of all the existing

chemicals or keep pace with those being developed

Current testing paradigm does not incorporate advances in

technology or biological understanding


ToxCast/Tox21 are U.S. Programs Using In Vitro

HTS to Address these Challenges

Tox21 ToxCast

• Began in 2005 with National Toxicology

Program and National Chemical Genomics

Center

• The U.S. Environmental Protection Agency

joined in 2006

• Five year MOU signed Feb 2008

• The U.S. Food and Drug Administration

joined in 2010

• Initiated inside a newly formed U.S.

Environmental Protection Agency Center –

National Center for Computational

Toxicology (NCCT)

• Began as a phased effort in 2007

• Planned Phase I 2008 - 2009

• Planned Phase II 2009 – 2010 (assay data

released by September, 2013)


What is In Vitro High-Throughput Screening and

How is it Being Applied to Toxicology?

• Use robotics to screen compounds in 96, 384, or

1536-well plates

• Screening assays are either cell-free (i.e.,

biochemical), cell-based, or use lower-order

organisms (e.g., zebrafish)

• Generally use re-purposed drug discovery

assays

• Performed in concentration-response format

called qHTS (quantitative HTS)

• Used as a broad survey of potential proximal

biochemical and cellular targets for chemicals

HTS

Throughput

Molecular mechanism

Immediate organismal relevance

10’s/day

1000’s/day

10,000’s/day

100,000’s/day

“MTS”“LTS”


Chemicals in the ToxCast and Tox21 Libraries • ToxCast Phase I library – 309 unique

chemicals

• ToxCast Phase II library – 776 unique

chemicals

• The ToxCast Phase I chemicals are

primarily pesticides with a small

number of industrial chemicals

• ToxCast Phase II are pesticides,

industrial chemicals, and failed drugs

• Tox21 Phase I library – 2737

chemicals

• Tox21 Phase II library – 8193 unique

chemicals

• Compounds particularly relevant to

food and nutrition

• 160 FDA GRAS substances

• 905 FDA EAFUS substances

Kavlock et al., Chem Res Toxicol., 2012

NCCT Presentation, May 14, 2009


Assays in the ToxCast and Tox21 Portfolio

• ToxCast Phase I – 667 in vitro

assays across 10 platforms

• ToxCast Phase II – ~700 in vitro

assays across 13 platforms

• The ToxCast assays cover 327

gene targets with at least one gene

in 293 out of 592 KEGG pathways

• The ToxCast assays are primarily

re-purposed pharmaceutical assays

run by commercial vendors

• Tox21 – ~50-100 in vitro assays

• The Tox21 in vitro assays are

primarily cell-based reporter and

cytotox assays run at the National

Chemical Genomics Center

Kavlock et al., Chem Res Toxicol., 2012

Tox21 Presentation, June 2010

Nuclear Receptor; 30; 25%

Genetox; 14; 12%

Cytotoxicity; 32; 27%

Signaling Pathway; 9; 8%

Apoptosis; 32; 27%Ion Channel; 1; 1%

Nuclear Receptor

Genetox

Cytotoxicity

Signaling Pathway

Apoptosis

Ion Channel


ToxCast and Tox21 Data Analysis and Release

ToxCast

• Release of data through multiple peer-

reviewed publications

• Many analysis related publications

Tox21

• Release of data primarily through on-line

repositories PubChem and CEBS

• Fewer analysis related publications


Analysis of the ToxCast In Vitro Screening Data

for Predicting Hazard

Model #1

... Model #84

60 In Vivo Endpoints

Chemical

Structural

Descriptors

~600 In Vitro High

Throughput

Screens

Broad-based Evaluation

of In Vitro-to-In Vivo

Predictive Performance

Thomas et al., Tox Sci., 2012

8 Classification Algorithms,

~12 Feature Selection, 84

Classification Model

Combinations

309 EPA ToxCast

Chemicals


Current High-Throughput In Vitro Assays Have

Limited Capability of Predicting In Vivo Hazard In Vitro Assays

Chemical Structurep = 0.0709

Range of

performance

across the 84

statistical

models

The ideal is a balanced

accuracy of 1

60 In Vivo Endpoints

0.5 is like flipping a

coin



But, In Vitro HTS Assays May Be Useful to Identify

Potential Molecular Initiating Events

In VivoAnimal Response

In Vitro

Assay

Response

Positive Negative

Positive A B

Negative C D

Odds Ratio = (A/B)/(C/D) Log2 Odds Ratios for All Significant In Vitro Assay: In Vivo Endpoint

Combinations

In Vivo Endpoint In Vitro Assay Odds

Ratio

p-value

Chronic Study, Rat

Acetylcholineaserase

Inhibition

Biochemical, Rat

Acetylcholinesterase Binding

87.0 < 0.0001

Biochemical, Human

Acetylcholinesterase Binding

60.6 < 0.0001

Biochemical, Human

Butyrylcholinesterase

Binding

12.8 0.0003

Biochemical, Bovine

Progesterone Receptor

Binding

9.6 0.0007

Chronic Study, Mouse

Liver Tumors

Cellular, Human Peroxisome

Proliferator Activated

Receptor Alpha Reporter

27.8 0.0021

Biochemical, Guinea Pig

Opioid Receptor, Kappa 1

Binding

22.4 0.0074

Biochemical, Human

Serotonin Transporter

Binding

22.4 0.0074

aThe in vitro assays for each in vivo endpoint were filtered to remove those

with odds ratios < 5 and p-values > 0.01.



Current Limitations of the ToxCast andTox21

Approach

• The current in vitro assays may not capture the biochemical and cellular

responses in the in vivo tissues with adequate fidelity (e.g., metabolic

competence, transporter expression , cell-cell interactions)

• The current suite of assays may not reliably capture context-specific

outcomes.

• The current set of assays may not provide sufficient coverage of pathways,

protein targets, and cell types.

• Significant cross-species differences between the in vitro assays and in vivo

endpoints being predicted.

• The current analysis does not capture assay efficacy (only potency)

• The number of positive chemicals present for each endpoint may not be

sufficient to redundantly capture the broad array of mechanisms leading to

in vivo toxicity.


An Example of Why Context is Important from a

Dietary Perspective



Dietary Perspective

13 men

Pre-diet

Low Fat

High Fiber Diet

Exercise

Prostate Cancer Cell Growth

Apoptosis

TP53 Levels



Dietary Perspective

Tymchuk et al. J Urology 2001

Ngo et al. Cancer Causes Cntrl 2002

In addition, TP53 protein was induced in the diet

and exercise group.

Does this mean that diet and exercise would be

considered toxic/hazardous?


A Project is in Progress to Provide Dietary Context

to the Tox21 and ToxCast Datasets

• Screening ~30 fruit and vegetable juices in concentration

response format across ToxCast and Tox21 in vitro assays • 6-8 concentrations with maximum of 10%

• Screen in a subset of cell and biological assays (Bioseek, ACEA,

Attagene, Novascreen assays)

• Follow protocol of Charles et al., Food Chem Toxicol (2002)


Screening Fruit and Vegetable Extracts

Charles et al., Food Chem Toxicol, 2002


A Project is in Progress to Provide Dietary Context

to the Tox21 and ToxCast Datasets

• Screening ~30 fruit and vegetable juices in concentration response

format across ToxCast and Tox21 in vitro assays • 6-8 concentrations with maximum of 10%

• Screen in a subset of cell and biological assays (Bioseek, ACEA, Attagene,

Novascreen assays)

• Follow protocol of Charles et al., Food Chem Toxicol (2002)

• The NTP Tox21 group is interested in collaborating and want to also

screen the juices across a subset of the NCGC assays

• The juices will be obtained from “organic” fruits and vegetables

obtained from the local supermarket, but will also be screened for

pesticide residues and heavy metals • Support provided by ILSI Technical Committee on Food and Chemical

Safety.


Acknowledgements My Lab

ACC-LRI

ILSI

Funding

Barbara Wetmore

Reetu Singh

Bethany Parks

Linda Pluta

Chad Deisenroth

Filipe Muhale

Joe Trask

Briana Foley

Michael Black

Eric Healy

Longlong Yang

Joshua Harrill

External Collaborators (cont)

Jay Zhao (EPA)

Ila Cote (EPA)

David Dix (EPA)

Daniel Rotroff (EPA)

John Wambaugh (EPA)

Richard Judson (EPA)

Keith Houck (EPA)

Bob Kavlock (EPA)

Matt Martin (EPA)

David Reif (EPA)

Cornelia Smith (CellzDirect)

Kimberly Freeman (CellzDirect)

Stephen Ferguson (CellzDirect)

Harvey Clewell

Mel Andersen

Mark Sochaski

Brittany Allen

Katherine Cantwell

Ed LeCluyse

Institute Collaborators

External Collaborators

Russ Wolfinger (SAS)

Lili Li (SAS)

Wenjun Bao (SAS)

Tzu-Ming Chu (SAS)

Bruce Allen (BAC)

Jeff Gift (EPA)

Woody Setzer (EPA)

Scott Wesselkamper (EPA)

Nina Wang (EPA)

Jason Lambert (EPA)

Janet Hess-Wilson (EPA)

Dan Petersen (EPA)

tox21 update: relevance of tox21 to nutrition - ilsi...

Documents