mining toxicity data to expand the domain of applicability...
TRANSCRIPT
Philipp MayerTechnical University of Denmark
Mining toxicity data to expand the domain of applicability of chemical activity
LRI-ECO 30
Length: 2015-2017
Budget: €150 000
Main ParticipantsARC (Lead) – Jon A. Arnot, James M. Armitage, Trevor Brown
UFZ – Beate I. Escher, Stefan Scholz, Annika Jahnke, Nils Klüver
DTU - Philipp Mayer, Stine N. Schmidt
THI – Barbara A. Wetmore*
DMER/TU – Don Mackay*
* Advisory role
+ CEFIC LRI Monitoring Team
Malyka Galay-Burgos
Todd Gouin
Joop Hermens
Mark Lampi
Paul Thomas
La50
Chemical activityµ = µ∗ + 𝑹𝑻 × 𝒍𝒏(𝒂)
Energetic state relative to pure liquid (0-1)
a = 0 : no activity
a = 1 : saturation for liquids
a < 1 : solids form crystals below 1
Proportional to Cfree (a=Cfree/SL) and fugacity (a=f/fL)
Diffusion & partitioning from high to low activity
Equal at equilibrium asediment = ainterstitial water = aworm (Di Toro et al, 1991)
Chemical activity - well established for water!Water activity (aw, 0-1) = relative humidity (RH, 0-100%)
“microbial fouling requires a certain aw”
http://wateractivity.com/education/basics-of-water-activity/
http://waterinfood.com
Baseline toxicity exerted at wide concentration range,but narrow chemical activity range
Reichenberg & Mayer, 2006, ET&C 25: 1239-1245.
In correspondence with:
“Ferguson Principle” (1939)
DiToro’s Target Lipid Model
Van Wezel’s critical membrane
concent. (40-160 mM)
Effective activity(Ea50, unitless)
0.000001
Tadpole Mouse Algea
0.00001
0.0001
0.001
0.01
0.1
1
Effective concentration(EC50, in M)
LRI-ECO30General Objective
• Further test & examine the chemical activity hypothesis for toxicity
and risk assessment
Methods/Approach
• Compile toxicity data & apply the chemical activity approach to a
series of relevant case studies
LRI-ECO30ECO30 Research Activities
Database Compilation
Physical-chemical
Properties
QA/QC
Toxicity Data
(ECs, MoA)
Chemical activity (a)
calculations
Categorization/Clustering
Analyses
Uncertainty
LRI-ECO30ECO30 Research ActivitiesToxicity Data (ECs, MoA)
1. In vivo, juvenile + adult, acute + chronic• Fish data (78,206 records, 3,032 chemicals) from 4,011 studies
• Mollusc and Crustacean data (39,955 records, 2,469 chemicals)
• Amphibian and Reptilian data (7,172 records, 554 chemicals)
• Invertebrates and other miscellaneous species data (21,117 records, 1,576 chemicals
2. Acute Fish Embryo Tests (FET) data
3. Chronic fish toxicity (Fish, Early Life Stage, FELST) data
4. Algal growth inhibition data
5. C. elegans (nematode)
A. fischeri (bacteria)
6. In vitro, bioassay (ToxCASTTM)
MoAExpert knowledge
Toxtree
From the bioassay itself
(in vitro)
9
1. In vivo, juvenile + adult, acute + chronic• Fish data (78,206 records, 3,032 chemicals) from 4,011 studies
• Mollusc and Crustacean data (39,955 records, 2,469 chemicals)
• Amphibian and Reptilian data (7,172 records, 554 chemicals)
• Invertebrates and other miscellaneous species data (21,117 records, 1,576 chemicals
Partner 1 - ARC
10Partner 1 - ARC
The ToxTest v1.0: Toxicokinetic Mass Balance Model
• Toxicokinetic (bioaccumulation) model for aquatic organisms (fish)
• Relates external water concentrations (e.g., LC50s) to internal concentrations
(CBR50s) and internal chemical activities (La50s)
• External chemical activity (i.e., CA in water phase) also provided as model
output to readily allow comparisons to internal CA
Ea50
Ea50 External Activity > Internal
Activity due to biotransformation?
(i.e. disequilibrium?)
11Partner 1 - ARC
KEY RESULTS
Disequilibrium factors (DF) for suspected baseline toxicants (Narc/Inert), chemicals with specific modes of
action (React/Spec) and chemicals which could not be confidently assigned to either category (Uncertain).
Whiskers = 1.5 IQR. NOTE: Biotransformation half-lives are predicted values based on available QSARs
DF = Ea50Water / Ea50Biota
12Partner 1 - ARC
SUMMARY
• Database consists of ~150,000 entries for >4,500 chemicals from >1,000 species - So far, most
data points categorized as “Not Assignable” are due to unconfirmed exposure concentrations
• Tentative MoA classifications for 2,510 fish acute lethal data entries: 1 - 982 Narcosis/Relatively
inert; 2 – 1,082 Reactive/Specific MoA; 3 - 446 Uncertain (Unknown/Unsure)
• Uncertainty in physical-chemical properties is an important consideration when applying the
chemical activity approach
• Biotransformation can lead to large differences between the chemical activity in water
(external) and in the organism – not always relevant though, as shown for Case Study
Exploring the chemical activity concept for in vitro data
13Partner 2
Define baseline chemical activity
for in vitro assays
• Translate the existing data on measured/modeled cellular
concentrations into chemical activity • Predict baseline chemical activity for
HTS reporter gene assays
Task in WP 3 Approach
Data mine HTS in vitro assays
• Select ToxCast and other in vitro assays that describe clearly defined
modes of action • Convert reported nominal
concentrations into chemical activity
Define chemical activity-based
Toxic Ratio (TRa) for in vitro assays
in relation to MoA
• Define TRa threshold for baseline toxicants
• Calculate TRa for specifically acting compounds
• Explore clustering and ranges of excess activity in relation to MoA
3.1
3.2
3.3
Define baseline chemical activity
EaB for in vitro assays
Measures/models
TRa =EaB
EaS
SL
Ea
SF
ECW ECw, CBR
Chemical activity-based Toxic Ratio
14Partner 2
• Adapting the mass balance model (Armitage 2014) to 384 and 1536
well plate format and parameterize with experimental data
fcell
=1
1+1
Kcellw
Vw
mcell
+K
FBSw
Kcellw
mFBS
mcell
+K
PSw
Kcellw
VPS
mcell
Exploring the chemical activity concept for in vitro data
Fischer, F., Henneberger, L., König, M., Bittermann, K., Linden, L., Goss, K.-U. and Escher, B. (2017) Modeling
exposure in the Tox21 in vitro bioassays. Chemical Research in Toxicology 30, 1197−1208.
15
Partner 2
Exploring the chemical activity concept for in vitro data
Fractions in cells fcell
with mass-balance model for partition coefficients
log Kow
0 2 4 6 8
che
mic
al f
ractio
n in c
om
pa
rtm
ent (%
)
0,01
0,1
1
10
100
fwater fmedium fcells
fcell
=1
1+K
mediumw
Kcellw
mmedium
mcell
+1
Kcellw
Vw
mcell
Modelled internal effect concentrations
in cells IECcell
are in similar range as IEC for algae,
daphnia and fish
algae
daphnia
fish
cell
1
10
100
1000
10000
IEC
(m
mo
l/kg
lip
) fo
r a
qu
atic s
pe
cie
s
an
d E
Ccell fo
r ce
lls
Escher and Schwarzenbach, 2002
Fischer, F., Henneberger, L., König, M., Bittermann, K., Linden, L., Goss, K.-U. and Escher, B. (2017) Modeling
exposure in the Tox21 in vitro bioassays. Chemical Research in Toxicology 30, 1197−1208.
Fischer, 2017
Define activity-based toxic ratios for in vitro assays in relation to mode of action
16Partner 2
• First step: rescale the mass balance model to 1536 well plate and
modeling the published data from Huang, R.L et al. (2011) and additional
new data from ToxCAST
• Second step: define baseline from unrelated cytotoxicity data (constant cellular membrane concentrations)
• Third step:
TRactivity
=activity
baseline
activityspecific MOA
=IEC
cytotoxicity
IECspecific MOA
PR
PPAR
p53ARE
0.001
0.01
0.1
1
10
100
1000
To
xic
ra
tio
TR
(re
late
d to
ce
ll c
on
ce
ntra
tio
ns)
17Schmidt and Mayer (2015) Chemosphere 120: 305-308
Partner 3
(1) Extending to polar and solid MOA 1 & 2 compounds- confirming the chemical activity range for baseline toxicity
Aruoja et al. (2011) Chemosphere 84: 1310-1320
Aruoja et al. (2014) Chemosphere 96: 23-32
-1 0 1 2 3 4 5
-5
-4
-3
-2
-1
0
MOA 1 liquid (n=46)
MOA 1 solid (n=4)
MOA 2 liquid (n=20)
MOA 2 solid (n=38)
Log Kow
Log
EC
50/S
L
-1 0 1 2 3 4 5
-2
-1
0
1
2
3
MOA 1 liquid (n=46)
MOA 1 solid (n=4)
MOA 2 liquid (n=20)
MOA 2 solid (n=38)
a=1 (SL)
a=0.1
Log Kow
Log
EC
50 (m
mol
L-1
)
(2) Extending to more compounds, MOAs and species- identifying and quantifying excess toxicity
All data from Fu et al. (2015):
• awaiting publication
Data selection:
• awaiting publication
Selected for analysis:
• awaiting publication
Fu et al. (2015) Chemosphere 120: 16-22
Figure removed,
awaiting publication
Conclusions• Transferring toxicity data to chemical activity:
1. Visually relative to regression for liquid solubility (very simple)
2. Conversions of e.g. LC50 to La50
• Both approaches are straight forward for a large group of neutral chemicals, but more challenging
for e.g. ionics
• Uncertainty/error of input data and model assumptions can be important
• Baseline toxicity at chemical activity 0.01-1, generally confirmed
• Toxicity at chemical activity << 0.01 shows excess toxicity
• More commonalities than differences between La50 and ILC50
Articles, published
1. Fischer, F.C.; Henneberger, L.; König, M.; Bittermann, K.; Linden, L.; Goss, K.U.; Escher, B.I. 2017. Modeling exposure in the Tox21 in vitro bioassays. Chemical Research in Toxicology 30, 1197–1208.
2. Mayer P & SN Schmidt. 2017. Comment on “Assessing Aromatic-Hydrocarbon Toxicity to Fish Early Life Stages Using Passive-Dosing Methods and Target-Lipid and Chemical-Activity Models” Environmental Science & Technology 51, 3584−85.
3. Stibany F, Schmidt SN, Schäffer A & P Mayer. 2017. Aquatic toxicity testing of liquid hydrophobic chemicals - Passive dosing exactly at the saturation limit. Chemosphere 167: 551-557.
4. Klüver, N.; Vogs, C.; Altenburger, R.; Escher, B. I.; Scholz, S., 2016. Development of a general baseline toxicity QSAR model for the fish embryo acute toxicity test. Chemosphere, 164, 164-173.
5. Thomas P, Mackay D, Mayer P, Arnot J and MG Burgos. 2016. Response to Comment on “Application of the Activity Framework for Assessing Aquatic Ecotoxicology Data for Organic Chemicals”. Environ. Sci. Technol. 50, 4141-4142.
Manuscripts
1. Scholz, S.; Duis, K.; Schreiber, R.; Lidzba, A.; Armitage, J.M.; Mayer, P.; Léonard, M.; Altenburger, R. Retrospective analysis of fish early life stage tests – association of toxic ratios and acute chronicratios with modes of action. Submitted.
2. Gobas FAPC, Mayer P, Parkerton TF, Burgess RM, van de Meent D & T Gouin. A Chemical ActivityApproach to Exposure and Risk Assessment of Chemicals. Minor revisions.
3. Hermens JLM, Cronin MTD, Escher BI, Mayer P, Roex EWM & P Thomas. Linking aquatic toxicitydata to chemical activity and target site concentrations - beyond non-polar narcosis. In revision.
4. Schmidt SN, Armitage JM, Arnot JA, Mackay D & P Mayer. Expanding the chemical activity domainfor algal growth inhibition tests for non-polar organic compounds. To be submitted.
5. Winding A, Modrzyński JM, Christensen JH, Brandt KK and P Mayer. Soil bacteria and protists showdifferent sensitivity to polycyclic aromatic hydrocarbons at controlled chemical activity. To be submitted.
6. Arnot, J.A.; Armitage, J.M.; Orazietti, A.; Gouin, T.; McCarty, L.S.; Mackay, D. Toxicokinetic evaluation of critical body residue and chemical activity data for fish. In preparation.
7. Various ECO.30 Project Authors. Exploring the merits and limitations of the chemical activity approach for chemical hazard and risk assessment. Planned.
Presentations (posters/platforms)
1. Armitage JM, Arnot JA, Orazietti A, Brown TN, Celsie
A, McCarty LS, Mackay D. 2017. Expanding the
evaluation of the chemical activity hypothesis for
toxicity assessment. SETAC Conference, Brussels,
Belgium.
2. Schmidt SN, Armitage JM, Arnot JA, Kusk KO, Mayer
P. 2016. Linking algal growth inhibition to chemical
activity: A tool for identifying excess toxicity. SETAC
Conference, Nantes, France.
3. Schmidt SN, Armitage JM, Arnot JA, Kusk KO, Mayer
P. 2015. Linking algal growth inhibition to chemical
activity. SETAC Conference, Salt Lake City, UT.
4. Armitage JM, Arnot JA. Mackay D. 2015. Why is
chemical activity successful as a metric of aquatic
toxicity? A gedanken experiment explains why.
SETAC Conference, Salt Lake City, UT.
5. Brown TN, Armitage JM, Arnot JA. 2015. Addressing
uncertainty in sub-cooled liquid property estimation:
Applications for chemical activity calculations.
SETAC Conference, Salt Lake City, UT.
THANK YOU!
24/11/2017LRI – PRESENTATION TITLE