LRE 1 URANIUM(VI) SPECIATION: MODELLING, UNCERTAINTY AND RELEVANCE TO BIOAVAILABILITY MODELS. APPLICATION TO URANIUM UPTAKE BY THE GILLS OF A FRESHWATER.

1

LRE

URANIUM(VI) SPECIATION:

MODELLING, UNCERTAINTY AND RELEVANCE TO BIOAVAILABILITY MODELS.

APPLICATION TO URANIUM UPTAKE BY THE GILLS OF A FRESHWATER BIVALVE

Frank Denison

University of Aix-Marseille 1Laboratory of Radioecology & Ecotoxicology

Laboratory of Radioecology & Ecotoxicology, IRSN/DEI/SECRE/LRE, Cadarache, France

2

LRE URANIUM: A FRESHWATER CONTAMINANT

Uranium is a widely distributed naturally occurring element

In oxic surface-waters uranium is predominantly found in the +6 oxidation state, as the UO2

2+ oxyion

Various industrial activities mainly related to the nuclear fuel cycle can result in environmental contamination

Uranium has a double toxicity: both radiological and chemical

To properly assess the impact of uranium contamination on the biota, factors that can modify its bioavailability and/or toxicity need to be accounted for

Factors that influence a metal’s bioavailability include both the physico-chemical characteristics of the exposure medium and biological factors such as the behaviour or physiological status of the exposed organisms

3

LRE MODELLING METAL BIOAVAILABILITY: A HISTORICAL PERSPECTIVE

In many studies the measured or modelled concentration of the free metal ion found to be correlated with toxicity or availability

Steady-state or equilibrium approaches to modelling metal – organism interactions, considering the equilibrium established between the exposure water and the biosurface, dominate the literature

Studies over the past 30 years have shown that a metal’s total concentration is a poor indicator of availability or toxicity

Various models built upon the equilibrium paradigm have been proposed: FIAM - Free Ion Activity Model

GSIM - Gill Surface Interaction ModelBLM - Biotic Ligand Model

This approach is analogous to Surface Complexation Modelling and therefore integrates easily with existing speciation models

4

LRE THE USE OF BIVALVES IN METAL – ORGANISM INTERACTION STUDIES

Bivalves are frequently used for biomonitoring studies, ideal for long duration monitoring due to:

• For benthic species, their location at the sediment-water interface exposes them to contamination from both sources

• Respiration and feeding by water ventilation ensuring a high throughput of environmental medium

Bivalves can respond to unfavourable conditions by reducing or stopping water ventilation (“clamming up”)

This behavioural response may give difficulties for the interpretation of accumulation studies over time scales where this phenomenon is significant

5

LRE

External solution

Internal solution

Cell interior

STEPS INVOLVED IN THE ACCUMULATION OF U(VI) BY BIVALVES

Internalised U(VI)

2

If ventilation rate varies as a function of water composition this will confound interpretation of U(VI) accumulation in terms of solution speciation

Ventilation rate

[UO2]T

Inhalant siphon

Exhalant siphon

UO22+

1

Transporter – U(VI) complex

Li (OH-, PO4, CO3)

UO2-

X

UO2Li

6

LRE

0

10

20

30

40

50

60

70

5.5 6.5pH

Va

lve

op

en

tim

e/ %

N = 94

N = 96

Effect of pH on valve opening (no U)

0

10

20

30

40

50

60

70

0 1 2[U] / µmol dm-3

Effect of uranium on valve opening

THE BEHAVIOUR OF THE ORGANISMS IS MODIFIED BY THE EXPOSURE MEDIUM

Objective of this study: Investigate effects of chemical speciation on uranium bioavailabilityTo minimise behavioural effects isolated gill tissues exposed

7

LRE PROCESSES INVOLVED IN METAL ACCUMULATION

1. Mass transport of metalfrom bulk solution

2. Metal’s speciation at biological interface

3. Formation of metal – transporter complex

4. Trans-membrane transport of metal

1.2.3.

4.

4.

8

LRE ASSUMPTIONS OF THE EQUILIBRIUM PARADIGM

1. Internalisation of metal rate-limiting step

SLOW FAST 2. Internalisation kinetics first order

Application of equilibrium based models requires measurement or prediction of the metal’s speciation

3. Metal’s speciation in the vicinity of the interface same as in bulk solution

4. Metal-transporter complex in equilibrium with dissolved metal species

5. No significant modification of the biological interface occurs

6. The activity of the involved transport systems is constant for all conditions

9

LRE URANIUM SPECIATION: STATE OF THE ART

Analytical techniques to directly measure the solution speciation of uranium not yet available for environmental concentrations

Thermodynamic equilibrium models used to predict speciation

Structural chemical model generally well known, however thermodynamic model constants uncertain: literature values of constants quite disperse for some species

A new database was compiled to meet the requirements of: Internal consistency Coherent to domain of application Traceability to original data sources Containing uncertainty estimations for all values

Data sources included:OECD-NEA, IUPAC and NIST databases, original articles

10

LRE URANIUM SPECIATION: STATE OF THE ART

Uranium(VI) has an extensive solution chemistry forming strong complexes with many ligands, both inorganic (OH-, CO3

2-, PO43-…)

and organic (EDTA, Citrate, NOM…)

Very significant changes to the distribution of U(VI) species occur on varying environmentally important solution composition parameters (e.g. pH, [CO3]T, [PO4]T)

11

LRE URANIUM SPECIATION: UNCERTAINTY

The modelled solution speciation of U(VI) is limited by the uncertainty of the thermodynamic constants

A computer program was written to perform uncertainty calculations by Monte Carlo analysis

12

LRE THE MODELLING PROCESS

The process of modelling involves the establishment of a relation between a natural system and the formal (model) system by the opposite processes of ENCODING and DECODING

N

NATURAL

SYSTEM

F

FORMAL

SYSTEM

ENCODING

DECODING

13

LRE THE MODELLING PROCESS

The process of encoding is a creative act and depends on a number of poorly defined factors including:

• What models have been successfully applied previously

• State of knowledge of processes involved in natural system

• Personal preference and scientific background of the modeller

• Experimental design:

• Input factors of natural system that are varied• Independence of input factors (interpretation of natural system output can be confounded if input factors are not varied independently)• Input parameter space investigated e.g. the chemical composition domain (may be subject to bias from preconceived ideas about the natural system)

14

LRE EXPERIMENTAL DESIGN FOR URANIUM ACCUMULATION EXPERIMENTS

The experimental design selected for performing the accumulation experiments was strongly influenced by the prevailing approaches to understanding metal – organism interactions:

• Accumulation is governed by the formation of metal – transporter complex(es)

• These complexes are in equilibrium with dissolved metal species

• Competition between U(VI) and other cationic species for the transporter binding site may occur

Although these preconceptions may bias subsequent model encoding, this is a valid approach to test the prevailing equilibrium paradigm of metal – organism interactions

15

LRE EXPERIMENTAL DESIGN FOR URANIUM ACCUMULATION EXPERIMENTS

Factors influencing the solution speciation of U(VI) were varied independently:

All experiments were performed at constant ionic strength (0.01)

Citrate concentration: 0 – 10 µM

Carbonate concentration: 10 µM – 10 mM

Phosphate concentration: 0 – 100 µM

Uranium concentration: 10 nM – 10 µMpH: 5 – 7.5

Concentrations of potentially competing cations were varied:

Calcium and Magnesium concentrations: 10 µM – 2.5 mMSodium and Potassium concentrations: 300 µM – 4.3 mMProton concentration: 30 nM – 10 µM

16

LRE U(VI) UPTAKE BY EXCISED GILLSEFFECT OF CITRATE

17

LRE U(VI) UPTAKE BY EXCISED GILLSEFFECT OF pH AND [UO2]T

18

LRE U(VI) UPTAKE BY EXCISED GILLSEFFECT OF pH AND [CO3]T

19

LRE U(VI) UPTAKE BY EXCISED GILLSEFFECT OF [Ca] AND [Mg]

20

LRE OVERVIEW OF RESULTS

The expected decrease in uranium uptake on increasing complexation was observed for citrate and carbonate (pH 7 & 7.5)

No significant change in uptake was observed on varying calcium and magnesium concentrations at constant ionic strengthThe results cannot be explained by a simple dependence on the free uranyl–ion concentration

However, increasing complexation by carbonate (pH 5 & 6) did not decrease uptake – opposite effect for carbonate suggesting accumulation of a carbonate species

Several hypotheses may be forwarded to explain the observed pH dependence: accumulation of U – OH species, H+ competition for binding sites, or non-competitive H+ inhibition

21

LRE MODELS TESTED

A number of different uptake models can be proposed based on the equilibrium paradigm involving:

• One or several metal species – transporter complex(es)

• One or several independent membrane transporters

• Competition for the transporter binding site(s) by H+

• Non–competitive inhibition of uptake by H+

A multi-hypotheses approach was adopted: a number of different models of increasing complexity were applied to the results

22

LRE

1. Internalisation of metal rate-limiting step

2. Internalisation kinetics first order

3. Metal’s speciation in the vicinity of the interface same as in bulk solution

4. Metal-transporter complex in equilibrium with dissolved metal species

5. No significant modification of the biological interface occurs

6. The activity of the involved transport systems is constant for all conditions1, 2 or 3 Uranium species considered to form

transporter complexes: UO22+, UO2OH+,

UO2(OH)20, UO2CO3

0, UO2HPO40

Single or multiple membrane transporters

Potential H+ competition for transporter site

62 MODELS

5. No significant modification of the biological interface occurs

NON-COMPETITIVE H+ INTERACTION

1, 2 or 3 Uranium species considered to form transporter complexes

Stability constant of metal species – transporter complex varies as a function of pH

10 MODELS

6. The activity of the involved transport systems is constant for all conditions

5. No significant modification of the biological interface occurs

NON-COMPETITIVE H+ INTERACTION

1, 2 or 3 Uranium species considered to form transporter complexes

Transporter kinetics vary as a function of pH

10 MODELS

MODELS TESTED

23

LRE CHEMICAL COMPOSITION SUB-DOMAINS

The chemical composition domain considered for the model fitting can influence the process of model encoding, potentially affecting both model selection and calibration

In order to assess the importance of this effect, the chemical composition domain investigated was divided into a number of sub-domains of increasing chemical complexity

Each model was then fitted to each chemical composition domain. The best-fit residual values were then tested against the chi-squared distribution, enabling the model hypothesis to be either rejected or retained at a defined probability

24

LRE •Summary of model fitting:

Increasing model complexity

Increa

sing

do

main

Adjustable parameters 1 2 2 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 6 6 6 6 6 6

Modelling performed with mean-value thermodynamic database

pH variable, PCO2=1, [Cit]=0pH variable, PCO2=1, [Cit] variablepH variable, PCO2 variable, [Cit]=0pH variable, PCO2 variable, [Cit] variable

passes 0.1passes 0.01fails 0.01

Modelling performed integrating thermodynamic database uncertainty

pH variable, PCO2=1, [Cit]=0pH variable, PCO2=1, [Cit] variablepH variable, PCO2 variable, [Cit]=0pH variable, PCO2 variable, [Cit] variable

% that pass 0.011 - 2525 - 50> 50

Adjustable parameters 1 2 2 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 6 6 6 6 6 6

Modelling performed with mean-value thermodynamic database

pH variable, PCO2=1, [Cit]=0pH variable, PCO2=1, [Cit] variablepH variable, PCO2 variable, [Cit]=0pH variable, PCO2 variable, [Cit] variable

passes 0.1passes 0.01fails 0.01

Modelling performed integrating thermodynamic database uncertainty

pH variable, PCO2=1, [Cit]=0pH variable, PCO2=1, [Cit] variablepH variable, PCO2 variable, [Cit]=0pH variable, PCO2 variable, [Cit] variable

% that pass 0.011 - 2525 - 50> 50

25

LRE CONCLUSIONS

Uranium uptake is strongly influenced by solution composition

Equilibrium – based models are successful in describing the system behaviour for relatively simple solution composition domains

However, as the chemical domain space increases, an increasing number of hypotheses can be falsified at a high confidence level

Although the equilibrium paradigm cannot be rejected as a hypothesis, the level of model complexity required to describe the observed behaviour significantly limits the utility of such an approach

Alternative modelling approaches (such as the non-competitive effects of H+ concentration presented) can be proposed to explain the observed uptake behaviour

26

LRE PERSPECTIVES

Thermodynamic constants used for predictive speciation modelling are uncertain

Input uncertainty propagation limits the predictive ability of speciation modelling. This needs to be considered in order to assess the applicability of this technique

The proper implementation of equilibrium – based bioavailability or toxicity models requires:

• consideration of speciation modelling uncertainty

• the testing of a large chemical composition domain space(correlation of free metal-ion concentration with

measured endpoint for a strong – ligand titration series is NOT sufficient evidence of equilibrium control)

27

LRE ACKNOWLEDGEMENTS

IRSN-LREJacqueline Garnier-Laplace Christelle Adam Jim Smith Claude Fortin Rodolphe Gilbin Marcel Morello Damien Tran Olivier Simon

Danielle Poncet-Bonnard Arnaud Martin-Garin Laureline Février Jan van der Lee Claudine Van Crasbeck Brigitte Ksas Virginie Camilleri Gaëla Grasset