R&D - SEAC Safety & Environmental Assurance Centre

Gellatly N., Clapp C., Cubberley R., Dhadra S., Glavin, S., Hadfield S., Jacquoilleot S., Jarman A., Jowsey I., Lovell S., Maxwell G., Mayne J., Moore C., Pendlington R., Pickles J., Reynolds, J., Saib O., Sheffield D., Simpson W., Stark R., Summerfield V., Tang D., Windebank S. and MacKay C. SEAC Unilever, Colworth Science Park, Sharnbrook, Bedford, MK44 1LQ, UK

Acknowledgements

Adapted from: MacKay et al. 2013. ALTEX 30 473-486

APPLICATION OF MODEL: 2,4-DINITROCHOLROBENZENE (DNCB) CASE STUDY

1

Estimation of bioavailability in skin - based on OECD TG 428 • DNCB & ex vivo human skin - Pendlington et al 2008 Cutan. Ocul. Toxicol. 27. 283-94.

• Penetration system modelled and partitioning and diffusion rates determined by fitting - Davies et al 2011 Toxicol. Sci. 119, 308-18

Estimation of reaction kinetics in skin • DNCB in solution & ex vivo human skin (tape stripped)

• Reaction-diffusion system modelled and reaction rate determined by fitting to experimental data

Assume ‘Direct Acting’ hypothesis (unaltered proteasomal processing) and determine properties of resulting peptides

Estimate average haptenated-pMHC surface density from considerations of:

1. the fraction of nucleophiles we expect to be haptenated

2. probability that a pMHC contains a haptenated nucleophile

Prediction tools

Proteasomal cleavage

(e.g. NetChop)

MHC I binding (e.g.

NetMHCpan)

average number of

pMHC generated

per protein

average

number of

nucleophiles

per pMHC

Modelling MHC class I processing & presentation 3 Is the nature (TCR affinity) of the antigen limiting?

• what kon/koff do TCRs have for cognate hapten pMHC?

Explore effect of pMHC surface density and kon/koff on probability of T-cell triggering with the available models (Zarnitsyna & Zhu, 2012). Simulations generated using ‘confinement time’ model of Dushek, et al, 2009. Figures from: Huppa & Davis, 2013; Aleksic et al., 2010

Predicting the threshold for T cell activation 4

Davies et al. 2011. Toxicol Sci. 119 308-18

Loss from skin

Loss from

formulation

Partitioning Diffusion

Modelling skin bioavailability and reactivity 2

Reactivity

Kin

etics

Skin

Bio

availa

bity

ex vivo human skin

ex vivo

human skin

Chemical and biological events driving induction of human skin sensitisation have been investigated for many years and are now relatively well understood at a qualitative level. Informed by previous efforts, we are developing a mathematical model of human CD8+ T cell responses following topical exposure to a sensitising chemical.

These studies are increasing our fundamental understanding of human skin responses to sensitisers, and with this approach we aim to move towards an ability to quantify the relationship between the dose per unit area of a sensitising chemical applied to the skin and the likelihood that an adverse outcome that will arise as a result.

There are three main components to this model – bioavailability, antigen processing/ presentation and CD8+ T cell response. The model is underpinned by quantitative information from scientific literature, with laboratory investigations and clinical research activities being undertaken where information is not currently available.

From an Adverse Outcome Pathway (AOP) to a mathematical model

1. Skin

Penetration

3-4. Haptenation:

covalent modification

of epidermal proteins

5-6. Activation of

epidermal

keratinocytes &

Dendritic cells

7. Presentation of

haptenated protein by

Dendritic cell resulting in

activation & proliferation

of specific T cells

8-11. Allergic Contact

Dermatitis: Epidermal

inflammation following re-

exposure to substance due

to T cell-mediated cell

death

2.Electrophilic

substance:

directly or via

auto-oxidation

or metabolism

Epidermis Epidermis

Lymph

Node

Induction Elicitation

Skin Allergy Adverse Outcome Pathway

Viable Skin

Stratum Corneum

Receptor Fluid

Vehicle

Part

itio

nin

g

Dif

fusio

n Dendritic cell

Lymphatic Vessel Proliferating

CD8+ T cell

Dendritic

cell

?

Lymph Node

Naïve

CD8+ T cell

Model Output

Dose of chemical applied to skin

Pro

b. O

f hapte

n-s

pecific

T c

ell

activation

chemical X

chemical

For all Unilever presentations see:

www.TT21C.org

Figure from Yewdell et al. 2003. Nat. Rev. Immunol. 3 952-61

Characterising human T lymphocyte responses to chemical allergen p-phenylenediamine (PPD)

6

CD4

CD8

Ki-

67

0µg/ml PPD 0.01 0.1

Allergen driven proliferation of total lymphocytes and individual T cell subsets measured by intracellular Ki-67 expression.

Amy Popple, Jason Williams,

Rebecca Dearman and

Ian Kimber

Adverse

Non-Adverse

allergic immune response

time

No. C

D8

+ T

ce

lls

dose Y

dose X

Use model human immune response prediction to inform risk assessment decision

Model output for DNCB human clinical exposures

Clinical benchmark data:

Friedmann 2007 Br. J. Dermatol. 157 (6) 1093-102 • Single exposure to DNCB applied over

volar forearm in acetone vehicle

• Sensitivity assessed by challenge with DNCB 4 weeks after application

• Data shows the proportion of the cohort sensitised for a given dose

5 Model evaluation and next steps

Next steps:

• Currently performing uncertainty analysis – collaboration with J. P. Gosling (Univ. Leeds, UK; funding from NC3Rs)

• Development of model-based risk assessment approach

Insights from these analyses will guide future model development (e.g. explicit modelling of T cell subtypes, formulation effects) and may provide human biomarkers suitable for verifying model prediction

Current model scope: antigen-specific CD8+ T cell response including

• naïve (N)

CD45RO-CD62L+or CD45RA+CD27+

• central memory (CM)

CD45RO+CD62L+ or CD45RA-CD27+

• effector memory (EM)

CD45RO+CD62L- or CD45RA-CD27-

• effector (E)

CD45RO-CD62L- or CD45RA+CD27-

Human, sensitiser-specific T cell data are largely unavailable:

• Make use of literature data e.g. response to infection

• Generate sensitiser-specific, human-relevant data

Further development: Modelling CD8+ T cell response

Sheeja Krishnan,

Grant Lythe and

Carmen Molina-París

Modelling the Skin Sensitisation Adverse Outcome Pathway (AOP) for Risk Assessment

SAFETY SCIENCE IN THE 21ST CENTURY For more information visit www.tt21c.org

DNCB human clinical sensitisation data from: Friedmann 2007 Br. J. Dermatol. 157 (6) 1093-102

• Single exposure to DNCB applied over volar forearm in acetone vehicle

• Sensitivity to DNCB assessed by challenge with DNCB 4 weeks after application

Apply exposure, protein reactivity and other biological information as model inputs

Simulation of exposure scenarios cited in Friedmann 2007

• Model output = predicted synapse formation rate

• Percentages marked correspond to reported human sensitisation rates