predicting estrogen receptor binding within categories

Predicting Estrogen Receptor Binding within Categories

Rick KolanczykEnvironmental Protection Agency

Office of Research and Development

National Health and Environmental Effects Research Laboratory

Mid-Continent Ecology Division

Duluth, MN

Estrogen Receptor Toxicity Pathway

• Need is derived from Food Quality Protection Act – “estrogens”

• Chemical binding to ER is known to cause adverse effects– Toxicity Pathway

• Diverse chemicals bind to ER

• Drug bioassays for ER binding are not relevant – drug-design assays optimized for high affinity binders– environmental hazard may come from low affinity binding

Development and use of a QSAR requires clear Problem Definition

• The purpose of the QSAR application must be well-defined (e.g., priority setting for testing, and chemical-specific risk assessment are two very different purposes)

• The chemicals of regulatory concern must be defined to develop a database of tested chemicals (training set) for QSAR development

EPA Office of Pesticide Programs:Food Use Inerts & Antimicrobials

– Get ‘lists’– Characterize structures– Assess coverage of existing data (training sets, TrSets)– Select chemicals for testing to strategically expand structure space

to maximize information gained from every structure tested

Defining the Problem: EPA Office of Pesticide Programs

Food Use Inerts List - Structures

List from OPP/RD included:

893 entries = 393 discrete chemicals + 500 non-discrete substances(44% discrete : 56% non-discrete)

393 discrete chemicals include:366 organics (93%)24 inorganics (6%)3 organometallics (1%)

500 non-discrete substances include:147 polymers of mixed chain length170 mixtures 183 undefined substances

Defining the Problem: EPA Office of Pesticide Programs

Antimicrobials/Sanitizers - Structures

Lists from OPP/AD included:

299 = 211 discrete chemicals + 88 non-discrete substances(71% discrete : 29% non-discrete)

211 discrete chemicals include:153 organics (72%) 52 inorganics (25%) 6 organometallics-acyclic (3%)

88 non-discrete substances include:25 polymers of mixed chain length35 mixtures 28 undefined substances

In-lab Testing = Training Set The Issue:Assay protocols optimized to increase confidence in quantifying activity of

low potency compoundsCompare EPA Office of Pesticide Programs chemical structures to existing

ER data

The assays:

1) ER binding –

• standard competitive binding assay optimized for low affinity binders

• Rainbow trout liver cytosol (rtER)

2) Gene Activation

• trout liver slice - vitellogenin mRNA

• Endogenous metabolism

Data Example – Pos. Binding; Pos. Gene Activation

rtER Gene ExpressionrtER Binding

CTRL0.00001

0.0001

0.001

0.01

0.1

1

10 P E2 control

PNMP

PNEP

PNPrP

PNBP

-10 -9 -8 -7 -6 -5 -4 -3 -2

Log Concentration (M)

Vtg

mR

NA

(F

ract

ion

of

max

imu

m E

2ef

fica

cy)

CTRL

0

10

20

30

40

50

60

70

80

90

100

110

120

-10 -9 -8 -7 -6 -5 -4 -3 -2 -1

E2

P

PNEP

PNMP

PNBP

PNPrPPIPrP

Log Concentration (M)

[3H

]-E

2 B

ind

ing

(%)

Building a Decision Tree with Categories

Chemical Inventory

Contains CycleYes

Non binder(RBA<0.00001)

No

ChemicalInventory

Contains CycleYes

Yes

Belongs to known Inactive

class


Alkylbenzsulfonic Acids• Sulfonic Acid Dyes• p-Alkyl Fluorobenzenes• Bis–anilines• Alkoxy Anilines• Imidazolidines• Isothiazolines• Alkyl Benzthiols• Pyrrolidiones

• Oxazoles• Benzamide• Furans• Sorbitans• Triazines

No

Alkylbenzsulfonic acids

-0.62 msrd

CH3 S

OH

O

O

3.56 calc.0.92 calc.

S

O

O

OH

S

O

O

OH

CH3

4.45 calc. 6.15 calc.

CH3

NH2

CH3CH3

NH2

CH3

CH3N

CH3

N

CH3

CH3

3.57 calc.

CH3

NH2

CH3

NH2

Bis-Anilines

O

CH3

O

OHOH

OH O

OH

O

O

CH3

O

OH

OH

3.15 calc. 5.89 calc.

Sorbitans Sulfonic acids dyes

0.61 calc. 2.45 msrd

O

S

N

CH3

O

NH

S

Isothiazolines

AYE -1.00 calc. RAC 0.08 calc.SYE -0.44 calc.

O-

SO

ON

N

NH

N

S

O-

OO

OH

O-

O

O-

SO

O

N

N

OH

SO

-

O

O

OH

N

N

O

CH3

S

O

O

OH

CH3

SO

O

OH

FUI AMI FUI

--0.86 calc. 0.39 calc.

O

CH3CH3

NOH

O

NH

O

CH3

CH3N

Br

O

N

Cl

Imidazolidines

-1.00 calc. -0.44 calc. 1.89 msrd

O-

SO

ON

N

NH

N

S

O-

OO

OH

O-

O

O-

SO

O

N

N

OH

SO

-

O

O

Cl

Cl

NH

O

NH

Cl

O Cl

SO

O

O-

ChemicalInventory

Contains CycleYes Contains 2 OH,

or OH and =0, at Spec. Dist

Yes

NoBelongs to known Inactive

class


• Alkylbenzsulfonic Acids• Sulfonic Acid Dyes• p-Alkyl Fluorobenzenes• Bis–anilines• Alkoxy Anilines• Imidazolidines• Isothiazolines• Alkyl Benzthiols• Pyrrolidiones


No

R 394

E 353 H 524

BBAA

Schematic representation ofSchematic representation of ER binding pocket (LBD), with ER binding pocket (LBD), with 3 sites of interaction shown 3 sites of interaction shown (A, B, C), and ER protein a.a. (A, B, C), and ER protein a.a. involved in the interactions involved in the interactions which chemical ligands.which chemical ligands. CC

T 347

CC

J. Katzenellenbogen

R 394

E 353 H 524

CC

T 347

HOOH

CH3 H

H H

H

AA BB

A_B Interaction A_B Interaction

Distance = 10.8 for 17-Estradiol

ChemicalInventory



Possible High Affinity,“A-B”; “A-C”; or

“A-B-C” type binder

Contains some attenuating feature

steric?; other?

High Binding Affinity “A-B”;“A-C” or “A-B-C” type

No

Non binder Ex: ProgesteroneCorticossterone(RBA<0.00001)

Low Affinity Binder“A-B”,“A-C” or “A-B-C” type

Assess strength of attenuation steric?;

other?

Some

Complete

VYes

Yes

No


class



Yes


No

No

ChemicalInventory






steric?; other?


No


Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site

V

Possible Low AffinityBinder

YesNo

Yes

No


class



Yes


Yes

No

Yes

0.000001

0.00001

0.0001

0.001

0.01

0.1

0 1 2 3 4 5 6 7 8

LogKow

Lo

gR

BA

Additional 46 chemicals tested as non-binders with

LogKOW below 1.3

ChemicalInventory






steric?; other?


No


Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site

Type “A”Contains Phenol

Fragment

V


No

YesNo

II

Yes

No


class



Yes


Yes

No

Yes

R 394

E 353 H 524

CC

T 347

HOAA BB

““A_site only” Interaction A_site only” Interaction

CH3

ChemicalInventory






steric?; other?


No


Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site


Fragment

V


• Alkyl Phenols • Alkoxy Phenols• Parabens• Salicylates

Belongs to known Active

class

Potency Rules / Equations

No

YesNo

II

Needs testing

Yes

No

No

Yes


class

• Fully-hindered Alkyl Phenols

Belong to untested classwith possible

Type A low affinity

Yes


• Mixed Phenols

No

Yes

No


class



Yes


Yes

No

Yes

3.20 msrd

OH

CH3

5.16 calc.

OH

CH3

CH3

CH3

CH3

CH3

Alkylphenols

3.57 msrd3.40 calc. 1.96 msrd

O

OH

O CH3

O

OH

O

CH3

CH3

O

OH

O

CH3

OH

O

O

CH3

5.43 calc.

Parabens

2.55 msrd 4.63 msrd 5.06 calc.

O

OH

O CH3

O

OH

O

CH3

O

O

CH3

OH

Salicylates

3.31 msrd

OH

CH3

CH3

CH3

4.06 msrd

OH

CH3

ChemicalInventory






steric?; other?


No


Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site


Fragment

V


Type “B”Contains identif. Type B

Fragment• Alkyl Phenols • Alkoxy Phenols• Parabens• Salicylates


class


No

YesNo

No

II

Needs testing

Yes

No

No

Yes

IIIBelongs to

known Inactive class



Type A low affinity

No

Yes


• Mixed Phenols

No

YesIII

No


class



Yes


Yes

No

Yes

R 394

E 353 H 524

CC

T 347

AA BB

““B_site only” Interaction B_site only” Interaction

H3C

NH2

ChemicalInventory






steric?; other?


No


Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site


Fragment

• Alkyl Anilines• Phthalates• Branched Phenones• Cyclo Phenones• p-subst Cyclohexanols• p-subst Cyclohexanones• ring subst (o-CH3, tri-CH3) Benzoates

V



Fragment

• N-alkyl Phenones• Non ring subst Benzoates



class



No

Yes No

No

II

Needs testing

Yes

No

No

Yes

Yes

Belongs to class with possible

Type B low affinity under investigation


class

IIIBelongs to




Type A low affinity

• ring subst p-CH3 Benzoates •Thiophosphate Esters• Mixed Organics• Cyclic Alcohols (not p-subst hexanols)• Cyclic Pentanones/Others• Br, I halobenzenes

No

No

Yes


class

Belong to untested class

Needs testing

Needs testing


• Mixed Phenols

No

No

No

Yes

Yes

YesIII


No


class



Yes


Yes

Yes

No

Yes

5.12 calc.

CH3

NH2

2.70 msrd

CH3

CH3

CH3

NH2NH2

CH3

1.96 msrd

Alkyl Anilines

2.47 msrd 6.82 msrd 2.83 msrd 4.91 msrd

O

O

O

CH3

CH3

O

CH3

CH3

O

O

O

CH3

O

CH3 O

O

O

O

CH3

O

O

O

CH3

O

CH3

Phthalates

3.07 calc.3.00 msrd2.73 msrd

CH3 CH3

CH3

OO

CH3

CH3

O

CH3

CH3

Branched Phenones

ChemicalInventory






steric?; other?


No


Special Rule Classes

Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site


Fragment


V



Fragment




class


No


No

Yes No

No

II

Needs testing

Yes

No

No

Yes

Yes




class

IIIBelongs to




Type A low affinity

• ring subst p-CH3 Benzoates•Thiophosphate Esters• Mixed Organics• Cyclic Alcohols (not p-subst hexanols)• Cyclic Pentanones/Others• Br, I halobenzenes

No

No

Yes


class


Needs testing

Needs testing


• Mixed Phenols

No

No

No

Yes

Yes

YesIII


No


class



IV

Yes


Yes

Yes

No

Yes

• DDT-Like• Tamoxifen-Like• Multi-Cyclohydrocarbons• Alkyl Chlorobenzenes

5.87 calc. 3.30 msrd

Multi-Cyclic Hydrocarbons

Cl

CH3

4.94 calc.

Cl

CH3

4.41 calc.3.88 calc. 5.47 calc.

Cl

CH3

Cl

CH3

Alkyl-Chlorobenzenes

Cl

O

N

CH3

CH3

OHO

OH

O

OH

O

OH

6.89 calc.4.03 msrdCH3

N

CH3O

CH3

Tamoxifen Like

6.53 calc.

ClCl

ClCl

Cl

DDT Like

Setting Priority for Further Testing: Food Use Inert and Antimicrobial Lists

Lower Priority if: -Chemical belongs to a class where testing showed no evidence of ER interaction (RBA < 0.00001); -LogP <1.3, or meets other class specific LogP cutoffs

General characteristics of these chemicals: -Acyclic (e.g., no benzene rings)-Cyclic but does not contain a likely H-bonding group;

RBA = relative binding affinity; (a ratio of measured chemical affinity for the ER relative to 17-beta-Estradiol = 100%)Log P = log of octanol/water partition coefficient (also known as Log Kow); is an indicator of lipophilicity

Setting Priority for Further Testing: Food Use Inert and Antimicrobial Lists

Higher Priority if: -Chemical belongs to class with evidence of ER interaction, and:-LogP > 1.3 (or other class-specific cutoffs)

General characteristics of these chemicals: -Contains at least one cycle (e.g., benzene ring);-Contains a possible H-bonding group;

Inactive Classes – Lower PriorityInactive Classes FI AMAcyclics 230 122Alkylbenzenesulfonic Acids 78 3Sulfonic Acid Dyes 9 1Sorbitans 7 0Monocyclic Hydrocarbons 7 0Cyclic caged Hydrocarbon 1 0Pyrrolidines 3 0Furans 3 0N-chain Phenones 1 0Oxazoles 0 3Triazines 1 13Isothiazolines 3 7Imidazolidines 0 8Cyclic Inorganics 0 3Fully-Hindered Alkylphenols 1 0Cyclic Pentanones/Others 2 0Inactive ranges in Otherwise Positive ClassesCyclic Hexanones of LogP<1.5 1 0Cyclic Hexanols of LogP<1.5 1 0Chlorobenzenes of LogP<4 1 0Inactives in Mixed Functional Grps/Heteratom ClassesMixed Phenols 4 1Mixed Organics 14 13Organometallics 0 1

Total Inactives 367 175

1.50 msrd 1.97 msrd 2.47 msrd 3.65 msrd 3.20 msrd

OHCH3 OH

CH3

OHOH

CH3

OH

CH3

Classwise Differences in Lower LogKow Cutoff

p-n-Alkyl Phenols

AlkylAnilines

NH2

0.90 msrd 3.05 msrd

NH2

CH3

1.96 msrd

NH2

CH3

2.40 msrd

CH3 NH2

1.39 msrd

CH3

NH2

p-n-Alkyl Anilines

Cl

2.84 msrd 4.41 calc. 3.88 calc.

Cl

CH3

Cl

CH3

Cl

CH3

4.94 calc.

p-n-Alkyl Chloro

benzenes

OH

1.23 meas

n-Alkyl Cyclo

hexanols

2.32 calc.

OH

CH3

3.37 calc.

OH

CH3

Active Classes – High PriorityActive Classes FI AMAlkyl Phenols 4 9Parabens 3 0Salicylates 1 0Phthalates 3 0Actives in Mixed Functional /Heteratom ClassesMixed Phenols 1 6Mixed Organics 1 0

Total Positives 13 15

Classes – Under InvestigationClasses FI AMMixed Phenols (?) 3 2Mixed Organics (?) 7 16Organometallics (?) 2 3Thiophosphate Esters (?) 1 0

Total Under Investigation 13 21

Inventory Status – September 20, 2007

Food Use Inventory Antimicrobials

393 Total Chemicals 211

367 (94%) Lower Priority 175 (82%)

13 ( 3%) Higher Priority 15 ( 7%)

13 ( 3%) Under Investigation 21 (10%)

Summary of Current FindingsFood Use Inert and Antimicrobial Lists

ChemicalInventory






steric?; other?


No


Special Rule Classes

Log Kow <1.3



other?

Some

Complete

Contains at least

one possible

H-bonding site


Fragment


V



Fragment




class


No


No

Yes No

No

II

Needs testing

Yes

No

No

Yes

Yes




class

IIIBelongs to




Type A low affinity

• ring subst p-CH3 Benzoates•Thiophosphate Esters• Mixed Organics• Cyclic Alcohols (not p-subst hexanols)• Cyclic Pentanones/Others• Br, I halobenzenes

No

No

Yes


class


Needs testing

Needs testing


• Mixed Phenols

No

No

No

Yes

Yes

YesIII


No


class



IV

Yes


Yes

Yes

No

Yes

• DDT-Like• Tamoxifen-Like• Multi-Cyclohydrocarbons• Alkyl Chlorobenzenes



Fragment



Yes




class• ring subst p-CH3 Benzoates•Thiophosphate Esters• Mixed Organics• Cyclic Alcohols (not p-subst hexanols)• Cyclic Pentanones/Others• Br, I halobenzenes

No


class

Needs testing

No

Yes

Yes

III


3.84 msrd 3.59 msrd 4.15 msrd 4.01 msrd 3.84 msrd

O

O

CH3

O

O CH3

CH3

O

O

CH3

O

O

O

O

Non Ring-subst. Benzoates

3.61 calc. 3.69 msrd 4.14 calc.

O

CH3

CH3

CH3

O CH3

O

OH

CH3

CH3

CH3

O

CH3

CH3

CH3

O

CH3

Ring-subst. (tri-CH3) Benzoates

2.64 msrd

O

CH3

O

CH3

O

CH3

O

CH3

2.75 msrd

Ring-subst. (o-CH3) Benzoates

O

CH3

O

CH3

2.75 msrd

Ring-subst. (p-CH3) Benzoates

2.47 msrd

O

O

O

CH3

O

CH3

Phthalate

Summary• The decision support system has been coded to perform

an automatic classification of discrete chemical inventories.

• Modifications to the decision tree are made as needed as new inventories are examined which contain new structure/activity classes

• Decision support system was designed to categorize chemicals with similar activity (ER binding potential) to determine common aspects of structure.

AcknowledgementsEPA Mid-Continent Ecology Divivsion-Duluth:

Pat Schmieder, Mark Tapper, Jeff Denny, Barb Sheedy, Mike Hornung, Hristo Aladjov

Student Services Contractors:Carlie Peck, Beth Nelson, Tori Wehinger, Ben Johnson,

Luke Toonen, Rebecca Maciewski, Will Backe, Melissa Dybvig, Megyn Mereness

International QSAR Foundation:Gil Veith

EPA Office of Pesticide Programs:Steve Bradbury, Jonathan Chen, Jean Holmes, Kerry Leifer,

Betty Shackleford, Najim Shamim, Deborah Smegal, Pauline Wagner

OASIS software (Laboratory of Mathematical Chemistry): Bourgas, Bulgaria – Ovanes Mekenyan

predicting estrogen receptor binding within categories

Documents

nondiscrete substances44

nondiscrete substances71

er protein

estrogen receptor binding

database of tested chemicals

existing er datathe

food use inerts antimicrobials

inactive classnon binder