Mechanistic bioaccumulation model(s) for ionogenic organic substances in fish

Transcription

1 LRI-ECO21 Mechanistic bioaccumulation model(s) for ionogenic organic substances in fish Cefic-LRI Annual workshop Brussels, Nov Partner 1 Partner 2 Partner 3 Partner 4 Partner 5

2 Rationale ECO21 1. Bioaccumulation data required for PBT & RISK assessment 2. Scarce empirical B data for ionogenic organic compounds (IOCs) 3. Majority of B models trained/developed for neutral organic chemicals 4. Modeling B for IOCs needs new input data + include ph dependency IOCs: Cationic surfactants Pharmaceuticals & Anionic surfactants veterinary drugs Halogenated Resin acids phenolics Biocides Naphthenic acids

3 ECO21 Influence of Hydrophobicity on Bioconcentration Neutral Organics Bioconcentration factor (BCF) can be predicted from K OW Rapid biotransformation reduces the BCF in comparison to K OW

4 IONOGENIC compounds ECO21 COOH OH SO3 SO4 Sulfonamide 1o amine 2o amine 3o amine 4o ammonium ACIDS BASES NEED: hydrophobicity and metabolization rates K OW BCF relation inadequate if IOC >95% ionized Susceptibility of IOCs to biotransformation?

5 BIONIC model (Armitage et al. 2013) ECO21 C C B 1 BCFSS WD k k T Can address ph-dependence of BCFs at gill: diffusive transport dominated by neutral form in tissue: ions mainly sorb to membrane (phospholipids) in liver: fish drug biotransformation rates human

6 IOC - BCF model* performance ECO21 78 BCF values for IONIZED acids 16 BCF values for IONIZED bases trioctylamine Problem 1: Scarce BCF data for >95% ionized IOCs Problem 2: Very limited applicability domain for QSARs for partitioning (membrane) and biotransformation rate due to few data * Armitage et al Environ. Toxicol. Chem. 32,

7 Objectives ECO21 1. Develop new experimental data; Membrane-water partition coefficient (K MW ) Intrinsic fish hepatic clearance rate (in vitro S9 test) 2. Explore development of new QSARs to parameterize B models for IOCs; Gill uptake rate: ph at gill surface Improve/adapt QSAR for K MW and k M for IOCs 3. Facilitate adaptation into existing screening tools; e.g., EPISuite - BCFBAF

8 Key IOCs tested ECO21 ~50 IOCs >99% charged 1 o amines sulfates 2 o amines sulfonates 3 o amines carboxylic acids 4 o amines phenolic acids ~20 IOCs with BCF data available (ECO21 goal) direct comparison of accuracy of BCF-model BUT, non-ideal chemicals to improve QSARs ~30 model IOC structures (ECO21 extension) understand influence of specific molecular features active filling of knowledge gaps systematic approach to expand applicability domains

9 Membrane-water partitioning K MW for ion =? ECO21 K OW works for neutral compounds K OW is a poor predictor for IOCs K MW data < Cations 50 Anions QSAR ECO IAM-HPLC >2016?

10 Membrane-water partitioning ECO21 With 2013 BIONIC v1 K MW -QSAR/assumptions: 12 IOCs predicted within ± 10 x 5 IOCs over-predicted x New QSARs in development

11 Biotransformation (k M ) ECO21 Current in vivo k M QSARs includes 10 ionized IOCs may get more in vivo k M IOC estimates with new BIONIC model Rainbow trout liver homogenate (S9): in vitro clearance Phase I & Phase II enzymes active extrapolate to in vivo k M intra- /inter- study consistency compare in vitro k M with in vivo k M

12 Biotransformation (k M ) in vivo QSAR IOCs >70% charged empirical data ECO21 in vivo 3 Cations 7 Anions QSAR ECO21 : 50 in vitro k M significant k M insignificant k M 13 9

13 ECO21 Intrinsic hepatic clearance (CL intr,in vitro ) In vitro data C 8 -(NH 3+ ) C 8 -(NH 2+ C) C 10 -(NH 3+ ) C 12 -(NH 3+ ) C 12 -(NH 2+ C) C 8 -(NH + (C)C) C 10 -(NH + (C)C) C 12 -(NH + (C)C) C 8 -(N + (C)(C)C) C 12 -(N + (C)(C)C) C 8 -SO 4 - C 12 -SO 4 - C 8 -SO 3 - C 12 -SO 3 - C 8 -CO 2 - C 10 -CO 2 - C 12 -CO 2 - does the ionic group influence k M? hydrophobicity? (Only tested up to C 12 ) position of the ionic group? branching (steric effects)? additional functional groups?

14 Gill uptake rate constant (k 1 ) ECO21 k 1 EW G W V E W = f (pk a, K OW,N ) G V = f (size, O 2, T) k 1 database for IOCs Paucity of empirical data (n ~ 100) Modeled k 1 typically within a factor of 3-5 for most IOCs (3 case studies) Poor performance for highly dissociated IOCs (e.g., LAS, pk a < 1) using original (2013) model Uptake of charged form explicitly considered (implemented in BIONIC V2.0)

15 LAS: Revised vs. original k 1 model ECO21 Overestimation Revised model 5x 5x Estimated log k 1 (L/kg/d) Original model LAS C10-2 C11-2 C11-5 C12-2 C12-3 C12-5 C13-2 C13-5 Underestimation Empirical log k 1 (L/kg/d)* * Tolls et al. 1997

16 Planned activity for Final Report: Improved k 1 + empirical (k M + K MW ) BCFs ECO21 Original Model (Armitage et al. 2013) BIONIC v2 Estimated model input parameters (K OW -based K MW, k M -QSAR) Empirical model input parameters K MW S9-based k M Predicted BCFs (L/kg) Predicted BCFs (L/kg) Measured BCFs (L/kg)

17 ECO21 continuation ECO21 Can in vitro measurements improve BCF predictions? - in progress Additional data gaps Full tissue data (phospholipids, plasma & structural proteins) specific factors (hydrophobicity) on IOC biotransformation rates High tier needs: measured K PLIPW, K plasma & in vivo k M (+ metabolites) Lower tier needs: improved QSARs + IVIVE expand the chemical domain for S9 QSAR: intrinsic clearance of IOCs, part of k M screening paradigm shortcutting uptake uncertainties expand chemical domain for K MW,ion -QSAR: move away from K OW dependent extrapolation

18 Thanks Questions?

19 Back-up slides

20 S9 vs. in vivo Connors et al 2013 ET&C on in vitro fish S9: baseline ability of uninduced fish to biotransform xenobiotics in an environmental exposure.

21 ECO21 BCF model: improved k 1 + empirical (k M + K MW ) C12-5-LAS pka < 1: >99.99% anionic in environment Gill uptake rate constant (k 1 ) k 1 = 58 L/kg/d (Revised) vs. 1.0 L/kg/d (2013) log K MW = 3.87 ± 0.35 Empirical BCF: L/kg* BCF Model: 6-20 L/kg Significant biotransformation CL int, in vitro 125 (96 153) ml/hr/g liver * Tolls et al. 1997

22 lo g C lip o s o m e (n m o l/k g ) Membrane-water partitioning LRI-ECO21 Solid supported membrane bilayer, easy to centrifuge and sample C free 96 well system problematic for surfactants adapted protocol: sample aqueous phase with LC-MS autosampler fa tty a m in e s S Negligible glass binding if >90% sorbed or, measure glass binding afterwards 7 Preliminary data 2015: 6 lo g K M W = 4 P 1 0 Niels Timmer, Steven Droge IRAS, Utrecht University lo g K M W T 1 0 = 3 Q lo g C a q,fre e (n M )

23 Biotransformation (k M ) Which metabolites are formed, do they inhibit k M? BIOTRANS- FORMATION dealkylation

24 IONOGENIC compounds IOC types REACH (> ): ~7% bases pk a >8 ~7% acids pk a <6 Drugs (~4000 API): ~40% bases pk a >8 ~10% acids pk a <6 Too many to measure Manallack 2007, Perspectives in Medicinal Chemistry : (1999 World Drug Index database lists compounds Franco et al Int. J. Life Cycle Assessm.,

25 In vitro clearance BIOTRANS- FORMATION Nice for empirical input in the IOC-BCF model, but how to apply? to in vivo QSAR-predictions that are poorly validated for IOCs? cross examine in vivo & in vitro transformation data to related IOCs? Understand S9 transformation capacities what is the role of the ionic group? identify future research needs

26 L o g c o n c e n tr a tio n ( M ) Biotransformation (k M ) BIOTRANS- FORMATION Does the ionic group influence k M? YES C 8 N H 3 + C 8 N H 2 C C 8 N H + (C ) C C 8 N + (C ) (C ) C C 8 S O 3 - C 8 S O 4 - C 8 C O O - T im e (m in ) Rapid clearance 3 o amines & sulfates

27 L o g c o n c e n tr a tio n ( M ) Biotransformation (k M ) BIOTRANS- FORMATION Does hydrophobicity influence k M? YES C 1 2 N H 3 + C 1 2 N H 2 + C C 1 2 N H + (C ) C C 1 2 N + (C ) (C ) C C 1 2 S O 3 - C 1 2 S O 4 - C 1 2 C O O - k M C12 > C8 T im e (m in ) Almost all IOC types transformed

28 L o g c o n c e n tr a tio n ( M ) Biotransformation (k M ) BIOTRANS- FORMATION Does steric hindrance influence k M? b e n z y l-n H + -( C ) C C 8 N H + (C ) C C 1 2 N H + (C ) C trib u ty la m in e T im e (m in ) No clearance benzyldimethylamine C8-NH + (C)C > tributylamine

29 Biotransformation (k M ) LRI-ECO21 Which neutral functional groups influence k M? = k M dicyclohexyl < dibenzyl < dihexyl (hydroph.?)