Modeling environmental impacts of engineered nanomaterials : the value of “generic models”

of individual organismsRoger M. Nisbet

University of California, Santa Barbara

Work with: Tin Klanjscek, Shannon Hanna, Trish Holden, Ben Martin, Ed McCauley, Bob Miller, Erik Muller, John Priester, Louise Stevenson, and many othersFunding: US Environmental Protection Agency and National Science Foundation (through UC CEIN).

The need for theory in ecotoxicology• Contaminants impact individual organisms, populations,

communities and ecosystems.

• Contaminants are one component of environmental stress, that typically acts simultaneously with others (e.g. temperature, pH, food availability……….)

• Gereral theory is required because testing cannot match rate of introduction of new chemicals:

- 75,000+ chemicals registered for commercial use in US- less than 1000 have undergone complete toxicity testing- overwhelming costs of tests ($2-$4 million for in vivo studies)

The need for theory in ecotoxicology• Contaminants impact individual organisms, populations,

communities and ecosystems.

• Contaminants are one component of environmental stress, that typically acts simultaneously with others (e.g. temperature, pH, food availability……….)

• Biology-based theory is required because testing cannot match rate of introduction of new chemicals:

- 75,000+ chemicals registered for commercial use in US- less than 1000 have undergone complete toxicity testing- overwhelming costs of tests ($2-$4 million for in vivo studies)

• Dynamics of budgets of energy and elemental matter should be a component of this theory.

• Kooijman’s DEB theory offers a powerful framework for this.

Definition Engineered nanomaterial (ENM) consists of intentionally produced particles with a characteristic dimension between 1 and 100nm and possessing properties that are not shared by non-nanoscale particles with the same chemical composition”

Examples- metal oxides – TiO2 and ZnO, (sunscreeen); Ag (antibacterial)- Quantum Dots (electronics)

Properties- Size and shape dependent due to: large surface/volume- Often manufactured with coatings

Ecological/environmental impact?- May impact biogeochemical fluxes (nutrient cycling)- Toxicity (e.g designed for antibacterial/antifungal properties)

Nanotechnology has made the challenge tougher

100’s/year 1000’s/year 10,000’s/day 100,000’s/day

High Throughput Bacterial, Cellular, Yeast, Embryo or Molecular Screening

Information on potential ENM hazard

Expensive in vivo testing and ecological experiments

few/year

Challenge for theorists: to use information from molecular and cellular studies to prioritize, guide design, and interpret ecological studies

Dynamic Energy Budget (DEB) Models

Organism

GrowthDevelopmentReproduction

Survival

Resources Metabolic Products

DEB model equations describe the kinetics of the “reactor” that converts resources into “products”

Kooijman’s “standard” DEB model

FecesJEA

ME

MV

somaticmaintenance

growth

k 1-kMaturity

Maintenance

MH

MER

Maturity orReproduction

JEC

Food

Reserve

Mobilization

X

Kooijman’s “standard” DEB model*

i-state variables Reserve biomass at time t Structural biomass at time t “Cumulative reproduction”, i.e. total carbon allocation to

reproduction buffer by time t Total allocation to “maturity” by time t . Hazard rate at time t, i.e. instantaneous “risk” of mortality Aging acceleration at time t – related to level of damage

inducing compounds Parameters

Total of ~12 parameters. Of these some are expected to be broadly invariant across taxa and others scale in predictable way with size. This opens the way to generality. For many applications, fewer state variables and parameters suffice.

S.A.L.M. Kooijman (2010) Dynamic Energy Budget models for metabolic organization. Cambridge University Press. T. Sousa et al (2010)., Philosophical Transactions of the Royal Society B, 365:3413-3428.

Kooijman’s “standard” DEB model equations

E EA ECd M J Jdt

( )V VG EC EM VEd M J J J ydt

k

(1 ) if , else 0pH EC EJ H H H

d dM J J M M Mdt dt

k

0 if , else (1 )pER H H ER EC EJ

d dM M M M J Jdt dt

k

2with ( ) { } if else 0 bEA EAm H H EAJ c T f J L M M J

2( ){ } 1EC EAmm

ge LJ c T J Lg e gL

3( )[ ]EM EMJ c T J L

( )EJ J HJ c T k M

PLUS ODEs for aging acceleration and hazard rates

Kooijman’s “standard” DEB model equations

E EA ECd M J Jdt

( )V VG EC EM VEd M J J J ydt

k

(1 ) if , else 0pH EC EJ H H H

d dM J J M M Mdt dt

k

0 if , else (1 )pER H H ER EC EJ

d dM M M M J Jdt dt

k

2with ( ) { } if else 0bEA EAm H H EAJ c T f J L M M J

2( ){ } 1EC EAmm

ge LJ c T J Lg e gL

3( )[ ]EM EMJ c T J L

( )EJ J HJ c T k M

3/ 8 6 2 ' ( VIRUDJLQJDFFHOHUDWLRQDQGKD]DUGUDWHV

COLLECTION OF MESSY

ODEs

Dynamics of structured populations

• Environment: E-state variables - experienced by all organisms - Resources - Toxicants - Metabolic products

• Individual Organism: i-state variables - DEB state variables – ODEs in previous slides

• Population dynamics: p-state variables – Book-keeping - population size, age structure, distribution of i-state variables - many mathematical representations possible (IBMs, PDEs, IDEs etc.) - special assumption (ontogenetic symmetry) yields ODEs

Population modeling involves assumptions on interactions of individuals and their environment

Messages from some UC CEIN Projects

1) Phytoplankton I. Ontogeny symmetry assumed. Suborganismal and population properties consistent

2) Phytoplankton II. Metabolic products important Algal-produced compounds mitigate toxicity.

3) Bacteria. Metabolic products important. Suborganismal data can help model selection.

4) Individual Population projection for mussels. Ontogeny asymmetry. Population response more sensitive than individual response 5) Phytoplankton-zooplankton interactions. Ontogeny

important and metabolic products important?

Effects of ENMs on phytoplankton

populations

Kooijman’s “standard” DEB model*

i-state variables Reserve biomass at time t

Structural biomass at time t “Cumulative reproduction”, i.e. total carbon allocation to reproduction

buffer by time t Total allocation to “maturity” by time t . Hazard rate at time t, i.e. instantaneous “risk” of mortality

Aging acceleration at time t – related to level of damage inducing compounds

Parameters Total of 3 parameters + 2 parameters for toxic effects. Of these some are expected to be broadly invariant across taxa and others scale in predictable way with size. This opens the way to generality. For many applications, fewer state variables and parameters suffice.

S.A.L.M. Kooijman (2010) Dynamic Energy Budget models for metabolic organization. Cambridge University Press. T. Sousa et al (2010)., Philosophical Transactions of the Royal Socitey B, 365:3413-3428.

Marine phytoplankton population growth*

• Study of 4 phytoplankton species exposed to TiO2 and ZnO particles

• No effect with TiO2

• ZnO effect probably due to Zn2+

Toxicity described by two quantities (NEC and one other)

* R.J. Miller et al. (2010) Environmental Science & Technology 44: 7329–7334

Marine phytoplankton population growth*

• Study of 4 phytoplankton species exposed to TiO2 and ZnO particles

• No effect with TiO2

• ZnO effect probably due to Zn2+

DEB modelToxicity described by two quantities (NEC and one other)

* R.J. Miller et al. (2010) Environmental Science & Technolgy 44: 7329–7334

Marine phytoplankton population growth*

• Study of 4 phytoplankton species exposed to TiO2 and ZnO particles

• No effect with TiO2

• ZnO effect probably due to Zn2+

DEB modelToxicity described by two quantities (NEC and one other)

* R.J. Miller et al. (2010) Environmental Science & Technology 44: 7329–7334

ZnO mg L-1 (ppm)

RF

ZnO mg L-1 (ppm)

Reactive oxygen species (ROS)production

Membrane permeability (Cell death )Mitochondrial membrane potential

ZnO mg L-1 (ppm)

Dynamic Energy Budget(DEB) modeling of NEC

NEC = 223 ± 56 ppb

Rel

ativ

e flu

ores

cenc

e (R

F)

Isochrysis galbana

Expt data from Cole, Cherr et al., in prep

18

Marine phytoplankton population growth*

BUT – it’s not always that simple(Expts by L. Stevenson on silver ENMs and a freshwater alga)

New culture One week old Two weeks old

Size of AgNPs (nm) Per capita growth rate of algal cultures5 mg/L citrate-coated AgNP

New culture One week old Two weeks old

Particles aggregate in older batch cultures

Smaller particles more toxic than aggregates

Hypothesis: algae excrete soluble organic compounds that interact with particles and dissolved metals ADDITIONAL FEEDBACK TERM(S) + NEW E-STATE INTERACTIONS

DOC mitigation of AgNP and Ag+

Effects of Cd-Se quantum dots on bacterial populations (Pseuomonas aerigunosa)

Strategy: Use DEB models to charcterize differences in bacterial growth response to Cd(II) and CdSe Quantum dot (QD) exposure

Contrasting QD toxicity with toxicity of dissolved Cd1-3

1. Data from J. Priester et al. Environmental Science and Technology 43:2589-2594 (2009).

2. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, PlosONE, 7(2): e26955. doi:10.1371/journal.pone.0026955)

3. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, Ecotoxicology, in review

Strategy: Use DEB models to charcterize differences in bacterial growth response to Cd(II) and CdSe Quantum dot (QD) exposure

Contrasting QD toxicity with toxicity of dissolved Cd1-3

1. Data from J. Priester et al. Environmental Science and Technology 43:2589-2594 (2009).

2. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, PlosONE, 7(2): e26955. doi:10.1371/journal.pone.0026955)

3. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, Ecotoxicology, in review

• New feedback to environment required to fit DEB model to control (zero Cd) curve

Kooijman’s “standard” DEB modeli-state variables Reserve biomass at time t

Structural biomass at time t

“Cumulative reproduction”, i.e. total carbon allocation to reproduction buffer by time t

Total allocation to “maturity” by time t . Hazard rate at time t, i.e. instantaneous “risk” of mortality

Aging acceleration at time t – related to level of damage inducing compounds

Acclimation energy – new variable

E-state variables Environmental degradation – new variable

Parameters Total of 6 DEB parameters + variable number of other parameters depending on submodel. Of these some are expected to be broadly invariant across taxa and others scale in predictable way with size. This opens the way to generality. For many applications, fewer state variables and parameters suffice.

Strategy: Use DEB models to charcterize differences in bacterial growth response to Cd(II) and CdSe Quantum dot (QD) exposure

Contrasting QD toxicity with toxicity of dissolved Cd1-3

1. Data from J. Priester et al. Environmental Science and Technology 43:2589-2594 (2009).

2. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, PlosONE, 7(2): e26955. doi:10.1371/journal.pone.0026955) (2012)

3. T. Klanjscek, J. Priester, P.A. Holden and R.M. Nisbet, Ecotoxicology DOI 10.1007/s10646-012-1028-7 (2013)

• New feedback to environment required to fit DEB model to control (zero Cd) curve

• Model with toxic effect on resource assimilation and mortality best fits response to Cd (II) and to ROS data

Modeling the effect of QDsRule of the game: no change in Cd toxicity model

QD dissolutionintroduces Cd2+ in environment

Cd2+ interferes with assimilation and enters the cell → previous toxicity model

QDs associate with the cell

Associated QDs produce ROS affecting membrane processes

ROS produced inside the cellaffect all cellular processes

CdSe

CdSe

CdSe

• Model selection from fitting growth trajectories not possible

• Measurements of Reactive Oxygen Species (ROS) allow model selection

Toxicity mechanism for Quantum Dots

Effects of metal oxide nanoparticles on populations of marine mussels (Mytilus spp.)

Adult marine mussels, Mytilus galloprovincialis, were exposed to ZnO NPs for 12 weeks at concentrations up to 2 mg L-1.

Basic measurements on individuals(2 food levels)1) weights of shell, gonad, somatic tissue2) Zn distribution within organism3) Tank clearance rates information on food consumed. 4) Iindividual clearance rates5) Oxygen consumption rates.

Population level predictionAims to extract enough information to project effects on lifetime reproduction

(previous experience in Muller, E.B. et al. Ecotoxicology 19: 38-47 (2010))

Effects of ZnO NPs on mussel physiology(Expts. By Shannon Hanna)

0 fecundity survival

at age to age

( ) ( )

a a

R a S a da

used to estimate parameters

From DEB model

EC50 EXPECTED LIFE-TIME PRODUCTION OF REPRODUCTIVE MATTER

- EC50 for a given food level

- MUCH SMALLER THAN FOR INDIVIDUAL RATES (e.g. 1.5 mg/l for feeding)

- Consequence of ontogenic asymmetry

Phytoplankton-zooplankton interactions

3

2

MaturationReproductionGrowth

4

Somatic maintenance

Maturity maintenance

Feeding

3

1

Reserve

3

2

MaturationReproductionGrowth

4

Somatic maintenance

Maturity maintenance

Feeding

3

1

Reserve

Physiological Mode of Action1.Reproduction (direct)2.Feeding3.Maintenance4.Growth costs5.Control

Phytoplankton-zooplankton interactions DEB-IBM predicts effects of ontogeny asymmetry*

* Unpublished work by Benjamin Martin

Xmax= 2*K020

4060

80100

120140

Xmax= 5*K

Xmax= 10*K0

2040

6080

100120

140

Xmax= 20*K

Mean length

ECx reproduction

% re

duct

ion

com

pare

d to

con

trol

Xmax= 2*K0

20406080

100120140160

Xmax= 5*K

Xmax= 10*K

0 25 50 75 90 950 25 50 75 90 950 25 50 75 90 950 25 50 75 90 95

020406080

100120140160

Xmax= 20*K

0 25 50 75 90 950 25 50 75 90 950 25 50 75 90 950 25 50 75 90 95

Maturation flux / Reproduction flux

Xmax= 2*K0

20

406080

100

120140

Xmax= 5*K

Xmax= 10*K020

4060

80100

120140

Xmax= 20*K

Abundance

ECx reproduction

% re

duct

ion

com

pare

d to

con

trol

Take home messages

1. Structured population models (or IBMs) can help relate sub-organismal information (cheap and fast) to population dynamics (slow, expensive and important)

2. Abstract representation of individual organism (Kooijman’s DEB theory) has practical value

3. Experiments are revealing new feedbacks involving metabolic products

4. Ontogeny asymmetry impacts levels at which toxic effects impact populations