chemical reactivity measurement and the predictive ... · of non-animal test development ......

Preface

This is the 64th Report of a series of workshopsorganised by the European Centre for the Valida -tion of Alternative Methods (ECVAM).

The main objective of ECVAM, as defined in 1993by its Scientific Advisory Committee (ESAC), is topromote the scientific and regulatory acceptance ofalternative methods, which have scientific rele-vance and which reduce, refine or replace the use oflaboratory animals.

One of the first priorities set by ECVAM was theimplementation of procedures that would enable itto become well informed about the state-of-the-artof non-animal test development and validation, andthe opportunities for the possible incorporation ofalternative methods into regulatory procedures. Itwas decided that this would be best achievedthrough a programme of ECVAM workshops, eachaddressing a specific topic, at which selected groupsof independent international experts would reviewthe current status of various types of in vitro testsand their potential uses, and make recommenda-tions about the best way forward.

A workshop on Chemical Reactivity Measurementand the Predictive Identification of Skin Sensitisers

was held at ECVAM on 23–25 May 2007, under thechairmanship of Frank Gerberick. The workshopwas attended by experts from academia, nationalorganisations, and industries. The aim of the work-shop was to review the state-of-the-art of methodsfor the identification of skin sensitisers based onmeasurements of chemical reactivity. Furthermore,consideration was given as to how such methodscould contribute to integrated testing strategies forthe eventual replacement of in vivo testing.

A number of recommendations listed at the endof the report are intended to promote the progressof relevant and reliable methods toward prevalida-tion and validation.

Definitions

In chemico: a term which refers to the use of abi-otic chemical reactivity methods as replacementsfor animal (in vivo) assays; its use is analogous tothat of in vitro (for cellular bioassays) and in silico(for computer predictions).Hapten: a low molecular weight chemical that isimmunogenic only when attached to a carrier pro-tein.

Chemical Reactivity Measurement and the PredictiveIdentification of Skin Sensitisers

The Report and Recommendations of ECVAM Workshop 64a

Frank Gerberick,1 Maja Aleksic,2 David Basketter,3 Silvia Casati,4 Ann-Therese Karlberg,5 PetraKern,6 Ian Kimber,7 Jean Pierre Lepoittevin,8 Andreas Natsch,9 Jean Marc Ovigne,10 CostanzaRovida,4 Hitoshi Sakaguchi11 and Terry Schultz12

1Procter & Gamble Company, Miami Valley Innovation Center, Cincinnati, OH, USA; 2Safety andEnvironmental Assurance Centre, Unilever, Colworth Science Park, Sharnbrook, Bedfordshire, UK; 3StJohn’s Institute of Dermatology, St Thomas’ Hospital, London, UK; 4ECVAM, IHCP, European CommissionJoint Research Centre, Ispra, Italy; 5Dermatochemistry and Skin Allergy, Department of Chemistry,Göteborg University, Göteborg, Sweden; 6Procter & Gamble Company, P&G Eurocor, Strombeek-Bever,Belgium; 7Faculty of Life Sciences, University of Manchester, Manchester, UK; 8Laboratoire deDermatochimie, University Louis Pasteur, Strasbourg, France; 9Givaudan Schweiz AG, Dübendorf,Switzerland; 10L’Oréal, Aulnay-sous-Bois, France; 11Global R&D, Safety and Microbial Kao Corporation,Haga-Gun, Japan; 12Biological Activity Testing & Modeling, Laboratory College of Veterinary Medicine,Knoxville, TN, USA

ATLA 36, 215–242, 2008 215

Address for correspondence: Frank Gerberick, Procter & Gamble Company, Miami Valley Laboratories, P.O. Box 538707,Cincinnati, OH 45253, USA.E-mail: [email protected]

Requests for reprints: Silvia Casati, ECVAM, IHCP, European Commission Joint Research Centre, Via E. Fermi, 21027Ispra (VA), Italy.E-mail: [email protected] document represents the agreed report of the participants as individual scientists.

Pre-hapten: a non-reactive sensitising molecule,transformed into a hapten by simple chemical trans -formation (air oxidation) and without the require-ment for a specific enzymatic system. Pro-hapten: a non-reactive sensitising molecule,transformed into hapten by a specific enzymaticsystem. Adduct: a conjugate formed in an addition reac-tion.

Introduction

The regulatory background

The aim of regulatory toxicology is to identify haz-ardous chemicals, including those that have thepotential to cause skin sensitisation. In the Euro -pean Union (EU), and soon within the Glob allyHarmonised System (GHS), the approach to this iswell characterised. The relevant EU legislationincludes the Dangerous Substances Directive (DSD;Directive 67/548/EEC; 1) and the DangerousPreparations Directive (DPD; Council Directive1999/45/EC; 2).

With the recent adoption of the EU regulationson Registration, Evaluation, Authorisation andRestriction of Chemicals (the REACH system; 3),further emphasis has been placed on the use of themost up-to-date methods, as well as on ensuringthat decisions are made by using all the availabledata, with the minimum of additional animal test-ing.

The tests traditionally used for the identificationof chemicals possessing the intrinsic ability to causeskin sensitisation are the guinea-pig maximisationtest (GPMT; 4), the Buehler occluded patch test (5)and the local lymph node assay (LLNA; 6). The firsttwo of these use a combination of induction andelicitation phases in the guinea-pig, where theextent of the induction of sensitisation is deter-mined as a function of the (erythematous) responseto topical challenge. In contrast, the LLNA quanti-fies the induction response in mice, by measuringproliferation in the lymph nodes draining the site oftopical application. The capacity of these methodsto identify skin sensitisation hazards has only beenformally validated for the LLNA (7–10). However,both within this validation process and via the pub-lication of other datasets, the guinea-pig methodsare also recognised to be of sufficient sensitivity andspecificity (11–13).

For the purposes of hazard identification, skinsensitisation assays are interpreted in the samemanner. In simple terms, if the results in the LLNAare positive (the stimulation of proliferation in testgroup lymph nodes is at least three times greaterthan in concurrent vehicle-treated controls), or if,at challenge, ≥ 30% of the guinea-pigs are positive

in a maximisation test or if ≥ 15% of the guinea-pigs are positive in the Buehler test, then the sub-stance is regarded as a skin sensitiser. It can thenbe classified formally and labelled according to theEU system as “R43: May cause sensitisation by skincontact”. Thus, labelling can be applied to chemicalsubstances exclusively on the basis of data from asingle animal test, human experience only beingtaken into account if it exists, and even then, nor-mally not being used to overturn positive animaldata (14, 15).

Ultimately, such a basic hazard identification isnot sufficient for the protection of human health,but merely represents the first step. Risk assess-ment and risk management are the processes thataim to deliver human health protection. To permitthis, the relative potency of a skin sensitising chem-ical is an absolute prerequisite. The measurementof skin sensitisation potency has been the subject ofmuch discussion in recent years, with expert groupsin the EU (16), in European industry (17) and theWorld Health Organisation (18) making closelysimilar recommendations. Essentially, they all rec-ommend that the optimal strategy is to use theLLNA to determine the EC3 value, i.e. the concen-tration of the test chemical needed to induce athree-fold increase in cell proliferation in the drain-ing lymph nodes. It is not appropriate here to gointo any detail of this procedure, as it is has beenthoroughly reviewed elsewhere (19). What is impor-tant is to appreciate its importance for characteris-ing skin sensitisation hazards and for facilitatingrisk assessment (20, 21). Thus, where the LLNA isreplaced by a non-animal strategy, then it becomesvital that the method(s) which form this strategyare also capable of delivering hazard characterisa-tion, i.e. an assessment of relative potency, as wellas basic hazard identification.

The contribution of chemical reactivity toalternative strategies for skin sensitisation

Alternative strategies for the predictive identifica-tion of chemicals which possess the potential tocause skin sensitisation, typically dissect the keycomponent parts of the sensitisation inductionprocess, and seek to make assessments/measure-ments which indicate whether a chemical has thecapacity to trigger each part of the pathway.Measurements can be made of cytokine responsesin keratinocytes, and of changes in the Langerhanscells (LCs) or their equivalents, associated withmigration/maturation, and so on. Within thisreview, the focus is on another key aspect — theability of a chemical to react with proteins and formstable (covalent) bonds. Ultimately, if a chemical isbioavailable and can trigger each part of the induc-tion pathway, then it will be a skin sensitiser.However, it may not be at all clear how the individ-

216 F. Gerberick et al.

ual measurements contribute to the potency of thesensitiser. This problem was considered by Jowseyand colleagues, who proposed a simple (but stillhypothetical) scheme, by which data from disparateassays could be combined to generate a reflection ofthe potency of a skin sensitiser (22). This approachdemonstrated how information on such aspects asepidermal bioavailability, protein reactivity, andkeratinocyte and LC responses, might be scaled upand combined to yield an index of sensitisingpotency. Of course, this could only be achieved inpractice if the individual components were mostlyavailable, but once they were, the substantial avail-able database of potency values would provide thenecessary resource of in vivo information (23).Additionally, when information from several assaysis combined in the manner proposed, it would haveto be recognised that inaccuracies in terms of sensi-tivity, but especially in terms of specificity, mightbecome exaggerated. Thus, in developing and test-ing the type of chemical reactivity assay discussedin this review, particular attention should be paidto ensuring that specificity is high.

Finally, on the subject of the utility of chemicalreactivity data in relation to the development of anappreciation of the potency of a skin sensitiser, it isimportant to draw attention to one special topic.The measurement of the reactivities of a range ofchemicals already demonstrates that, unsurpris-ingly, the values vary widely (24–26). Reactivity isalso a key parameter in many QuantitativeStructure–Activity Relationships (QSARs) for skinsensitisation (27, 28). However, here, it is impor-tant to recognise that the reactivity measures cur-rently under evaluation are not intended to reflectthe specific mechanism(s) of protein haptenationrelevant to the induction of skin sensitisation invivo. Currently, it is not clear what specific aspectsof the reaction(s) might be relevant for potency,such as amino acid selectivity, reaction rate, andstability of protein conjugates. In addition, factorssuch as epidermal bioavailability (as opposed to thesolution stoichiometry of the reactivity assay)would be expected to influence sensitisationpotency. Thus, it is not appropriate to compare cur-rent reactivity measures directly with skin sensiti-sation potency measures such as LLNA EC3 values.

The Underlying Biological andChemical Mechanisms of SkinSensitisation

The immunobiology of skin sensitisation

Skin sensitisation leading to allergic contact der-matitis results from the topical exposure of aninherently susceptible individual to a chemicalallergen, such that a cutaneous immune response of

sufficient vigour is induced. If the now-sensitisedsubject is exposed subsequently to the same chemi-cal, at the same or at different skin site, then anaccelerated and more aggressive secondary immuneresponse will be provoked, causing a local inflam-matory response that is recognised clinically asallergic contact dermatitis (29).

The successful acquisition of skin sensitisationrequires complex cellular and molecular interac-tions that are tightly regulated in both time andspace (29, 30). It is necessary that the chemicalallergen is able to gain access across the stratumcorneum in order to reach the viable epidermis. It ishere that many of the pivotal events and processestake place, including the formation of stable associ-ations between the chemical allergen and pro-teins/glycoproteins. In immunological terms,chemical allergens are haptens, and as such areunable to elicit immune responses. For this to beachieved, they must bind with a protein to form animmunogenic complex. Such complexes interactwith epidermal LCs, and probably other cutaneousdendritic cells (DCs). Although there has recentlybeen some debate about whether, in all circum-stances, the elicitation of cutaneous immuneresponses to chemical allergens has a mandatoryrequirement for LCs (31), it is the case that, in mostinstances, it is these cells that orchestrate theevents which result in skin sensitisation. Followinga topical encounter with a skin sensitising chemical,LCs internalise and process the antigen which theytransport from the skin, via afferent lymphatics, tothe regional lymph nodes. During their movementfrom the skin, the LCs undergo a functional matu-ration, such that by the time of their localisationwithin lymph nodes, they are equipped to presentantigen very effectively to responsive T-lympho-cytes. The activation and clonal expansion of allergen-responsive T-lymphocytes, signals theacquis ition of sensitisation (32, 33).

The migration and functional maturation of LCsin response to skin sensitisation is regulated bycytokines and chemokines. It is now clear that themobilisation of LCs within the epidermis, and theirdirected movement from the skin to the regionallymph nodes, are dependent on the local availabilityof at least three cytokines that are native orinducible products of epidermal cells: interleukin(IL)-1β, IL-18 and tumour necrosis factor-α (TNF-α). The transport, processing and presentation ofan antigen is therefore dependent upon the induc-tion or up-regulation of these cytokines, and for thisreason, it is believed that, for the effective inductionof skin sensitisation, chemical allergens must causesome modest amount of trauma at the site of appli-cation, sufficient to drive the new or increased pro-duction of proinflammatory cytokines (30). Thecompletion of antigen transport also requires thatLCs trafficking from the skin, home to, and localisewithin, the appropriate site within the draining

ECVAM Workshop 64 217

lymph nodes (the paracortical region), where theycome into closest contact with T-lymphocytes. Tothis end, LCs are guided into lymph nodes bychemokines that are derived from cells already res-ident within nodes (30, 32, 34–38).

The central event in the acquisition of skin sensi-tisation is the activation of responsive T-lympho-cytes, which are stimulated to divide anddifferentiate. The division of the activated cellsresults in the selective clonal expansion of allergen-reactive T-lymphocytes that will respond in anaccelerated and more aggressive fashion followingsubsequent encounters with the same chemicalallergen (39).

Against this background, and in the context ofthis report, it is relevant to consider those eventsthat are regarded as essential for the successfulacquisition of skin sensitisation. The interestingquestion to pose, is why do not all chemicalsencountered at skin surfaces cause sensitisation?The reason is that, for a chemical to induce skinsensitisation, it has to clear a number of hurdles,and if any one of these is not negotiated success-fully, then sensitisation will fail to develop (or atleast develop optimally). To clear these biologicalhurdles, a chemical must:

— gain access to the viable epidermis across thestratum corneum, such that the necessary cellu-lar and molecular interactions can be initiated;

— form stable associations with proteins, in orderthan an immunogenic complex is created; thisrequires that the chemical is inherently protein-reactive, or can be transformed into a protein-reactive species within the skin;

— cause dermal trauma sufficient to provoke theinduced or up-regulated expression of thosecutaneous cytokines that are needed for theorchestration of LC mobilisation, migration andmaturation; and

— be inherently immunogenic, such that a T-lym-phocyte response of the necessary magnitudewill be induced.

It is therefore apparent that the acquisition of skinsensitisation is highly dependent upon complex andcoordinated chemical and biological events that cul-minate in the elicitation of a cutaneous immuneresponse, characterised by the activation and prolif-eration of T-lymphocytes in the regional lymphnodes.

The chemistry of skin sensitisation

It follows from the understanding of the biologicalmechanism, that for a chemical (hapten) to be a

sensitiser, it must have the ability to bind to a pro-tein so that a non-self, hapten–protein complexcan be produced. The hapten is usually a smallmolecule unable to raise an immunogenic responseby itself. Thus, immunogens are only formed ascomplexes with larger self-proteins. The chemicalreactivity of a hapten encompasses variousprocesses, which may include strong or weak inter-actions, such as covalent protein modification,non-covalent (reversible) interactions or oxido-reductive events.

Historically, irreversible covalent protein modi-fication has always been considered as the maindriver for the induction of sensitisation. This wasalready postulated in 1936 by Landsteiner andJacobs (40). The importance of non-covalentreversible hapten protein interactions, such ashydrophobic interactions or ionic bonds, has beendiscussed in the context of skin sensitisation.However, those interactions are not considered toplay a key role in determining the allergenicpotential of a chemical. Potential reaction mecha-nisms will be described in detail in the followingsection.

Moreover, it is estimated that about one-thirdof the skin sensitisers are not direct reacting hap-tens, but require some activation into a reactivespecies before they are able to bind to skin pro-teins. It was hypothesised many years ago, byDupuis and Benezra (41), that such molecules(termed pro-haptens) are actively metabolised bycutaneous enzymes into protein reactive species.Additionally, some other chemicals (termed pre-haptens) can also be chemically activated into pro-tein reactive species within the skin or prior to theabsorption, as a result of interaction with envi-ronment (e.g. by air oxidation, bacterial degrada-tion on the skin surface, or by photoactivation).

Molecular mechanisms of hapten–proteininteractions

The nature of chemical interactions

Haptens (generally small molecules with a molecu-lar weight less than 1000Da) can interact with bio-logical macromolecules by various mechanisms,leading to the formation of bonds of variousstrengths (Figure 1).

These chemical bonds are the result of electronicinteractions between atoms, and are characterisedby the energy involved. This reflects the bond sta-bility, and in general, a distinction is made betweenso-called “weak interactions”, involving energy lev-els from a few Joules to around 50kJ/mol, and so-called “strong interactions”, covalent orcoordination bonds, with energies ranging from 200to 450kJ/mol.


Weak interactions

Weak interactions are often grouped into threemain categories: hydrophobic bonds (including vander Waals’ interactions), dipolar bonds (includinghydrogen bonding), and ionic bonds (based on elec-trostatic interactions). Although these weak inter-actions involve modest energy levels and producestructures of low stability, they are nonetheless ofgreat biological importance, as they control all thephenomena of interaction between receptors andsubstrates and may play a role in the tissuebioavailability of the chemical.

Hydrophobic bonds represent the ability oforganic molecules to organise themselves in waterso as to minimise the contact area that they exposeto the aqueous solvent. It is, for example, by suchmeans that hydrophobic molecules insert them-selves into the phospholipid bilayers of cell mem-branes and into the hydrophobic regions ofproteins or membrane receptors. These hydropho-bic bonds, which involve energies of the order of40–80J/Å2/mol, could nevertheless play a role inallergies to very lipophilic products (42), such asthe allergens from poison ivy (Rhus radicans) orpoison oak (Rhus diversiloba). This could also beof importance for the interaction of haptens withthe lipophilic domains of antigen-presenting cells(APCs).

Dipolar bonds are electrostatic interactionsbetween permanent or induced dipoles. The elec-tron clouds do not always have a uniform chargedensity, and zones of high electron density caninteract electrostatically with zones of low electrondensity (permanent dipoles), or they could also bedeformed and polarised as they approach oneanother, thus creating induced dipoles. Hydrogenbonding is a special case of dipolar interaction, andcan occur between a hydrogen atom, bound to anelectron-withdrawing atom, and an electron-richatom. The energy of such bonds can be as high as25kJ/mol.

Ionic bonds are based on electrostatic interac-tions between pre-existing and generally localisedcharges on organic molecules or minerals. Suchinteractions occur, for example, between thecharged amino acids in proteins, and are thereforeimportant in recognition phenomena.

Strong interactions

Covalent bonds result when two atoms share a pairof electrons. They involve energies of the order of200–450kJ/mol, and are therefore very stable com-pared to the weak interactions. The two electronsrequired for bond formation can be contributed byboth partners, which is called a radical reaction, orcan both be provided by one of the atoms which isespecially electron-rich, and shared with an elec-tron-poor atom; this case is referred to as a reactionbetween a nucleophile (electron-rich) and an elec-trophile (electron-poor) centre. These two terms,nucleophile and electrophile, represent the capacityof a molecule, or rather an atom or a group of atomsof this molecule, to donate or accept electrons toform a covalent bond. Nucleophilic centres aretherefore partially negatively charged, while elec-trophilic centres are partially positively charged.

Coordination bonds are another type of relativelystrong bond, comparable even to covalent bonds atthe lower energy range; these occur between metalsor metal salts and electron-rich atoms (mainly het-eroatoms, such as nitrogen or oxygen). These inter-actions allow electron-rich groups or ligands totransfer part of their electron density to the metaland increase its stability. Coordination bonds arecharacterised by the number of ligands and by ageometry characteristic both of the metal and of itsoxidation state. For example, cobalt(II) is charac-terised by a tetrahedral arrangement, nickel(II) bya square planar tetra-coordinated arrangement,and chromium(III) by a six-ligand octahedralarrangement. The number of ligands and the geom-etry of these coordination complexes determinewhether the metals are allergenic, and controlcross-reactions to other related materials.

Mechanisms of bond formation

Even though weak or non-covalent interactionsmay play a role in the activation of haptens, the for-mation of a covalent bond between the chemicalhapten and a skin protein is certainly the mostimportant reaction involved. Many mechanisms areproposed for the formation of covalent bonds, but inthe field of contact allergy, they can be grouped into

Figure 1: Energy levels of chemical bonds

0.5–8.0 20 50 200 400 kJ.mol–1

van der π–π H-bond Ionic bond Coordination CovalentWaals bond bond


three main categories, even if other possibilities canbe proposed: nucleophilic substitutions on either asaturated or unsaturated centre, and nucleophilicadditions (Figure 2).

Nucleophilic substitution on a saturated centreinvolves the attack by an electron-rich nucleophileon an electron-poor electrophilic centre. The overalleffect will be a substitution of one of the groups (theleaving group) by the nucleophile. Nucleophilic sub-stitution reactions can also take place at an unsatu-

rated centre. The presence of a multiple bond per-mits the formation of a saturated intermediate andthe subsequent reformation of the multiple bond,permitted by the departure of a leaving group,resulting in the substitution product. This mecha-nism is illustrated in the aromatic series in which itis all the more favoured by attracting groups (e.g.nitro groups), which stabilise the intermediate.Nucleophilic addition is simply the addition (withno leaving group) of a nucleophilic atom to an

Figure 2: Principal mechanisms of covalent bond formation in contact allergy

Example of nucleophilic substitution on a saturated centre:

Example of nucleophilic substitution on an unsaturated centre:

Example of nucleophilic addition:

Nu

+ H–Br :

O

Br

H–Nu :

O

H–Nu

NO2

F

–

+

¨

Nu–H

O

O

–

+

: Nu–H

O

O

Nu

O

O

H–Nu : NO2

–H–F :

F

NO2

Nu


unsaturated electrophilic centre (containing one ormore multiple bonds). This mechanism is very sim-ilar to the first stage of nucleophilic substitution onan unsaturated centre, but the absence of a leavinggroup rules out the reformation of the multiplebond. A saturated compound is thus produced.

The principal electrophilic chemical groupspresent in contact allergens

Many molecules have electrophilic properties, andare thus able to react with various nucleophiles toform covalent bonds. Figure 3 shows the chemicalfunctions most frequently found in contact aller-gens, and the mechanism by which they react withnucleophilic groups. The previously-defined threemain types of mechanism (Figure 2) can be seen: i)nucleophilic substitution on a saturated centre (e.g.alkyl halides and epoxides); ii) nucleophilic substi-tution on an unsaturated centre (e.g. aromatichalides or esters); and iii) nucleophilic addition (e.g.carbonyl derivatives and alpha-beta-unsaturatedsystems). It should be noted that these mechanismsare the most probable ones, but that more-complexreactions or multi-step reactions can take placewith some sensitisers; this will be illustrated laterin this report.

If biological systems are considered from a chem-ical viewpoint, it becomes apparent that a verylarge proportion of structures, especially nucleicacids and proteins, contain electron-rich groups(those containing nitrogen, phosphorus, oxygen, orsulphur). Thus, overall, biological systems can beconsidered as being nucleophilic. It is therefore notsurprising that many biological mechanisms aredisturbed on contact with electrophilic chemicalsubstances. Depending on the site of action of theseelectrophilic molecules, the effect can be mutagenic(43), toxic (44), or allergenic if the target is the epi-dermis. In proteins, the side chains of many aminoacids contain electron-rich groups capable of react-ing with allergens (Figure 4).

Lysine and cysteine are those most often cited asexamples, but other amino acids containing nucle-ophilic heteroatoms, such as arginine, histidine,methionine and tyrosine, can react with elec-trophiles (45). Thus, it has been shown in nuclearmagnetic resonance (NMR) studies that nickel sul-phate, for example, interacted with the histidineresidues of peptides (46), and that methyl alkane-sulphonates, which are allergenic methylatingagents, mainly reacted with histidine and to a lesserextent with cysteine, lysine, methionine and tyro-sine (47). If the chemical structures of some aller-gens (Figure 5) are considered in the light of thechemical principles already outlined, it is easy todeduce that all of these molecules are be able toreact with biological nucleophiles.

The extremely stable covalent bonds formed in

this way could then lead to the triggering of delayedhypersensitivity. Again, the three main types ofmechanism for the formation of covalent bonds pre-viously described, are seen; the arrows indicate thereactive centres of each molecule (Figure 5).

The chemical selectivity of haptens for amino acids

A direct consequence of the diversity of hapten–pro-tein interactions is the existence of selectivity foramino acid modifications. For example, it wasshown that the alpha-methylene-gamma-butyrolac-tones, the major allergens of plants of theAsteraceae family, principally modify lysineresidues when reacted with human serum albumin(HSA; 48). It has also been shown that not all themodifications were antigenic, and that the sensiti-sation potential of a molecule is probably morerelated to its ability to modify some specificresidues, rather than to the modification of a largenumber of amino acids. Thus, the difference in sen-sitising potential of two sultone derivatives, analkenylsultone (a strong sensitiser) and an alkylsul-tone (a weak sensitiser), which differ only by thepresence of a double bond, could be better explainedby the selective modification of lysine residues bythe strong sensitiser than by the many tyrosineresidues modified by both derivatives (49, 50). Thesame observation can apply to 5-chloro-2-methylisothiazol-3-one (MCI) and 2-methylisothia-zol-3-one (MI), the main components of Kathon CG.While both these molecules were very reactivetoward cysteine residues (51), the strong sensitiser,MCI, was also shown to react with lysine and histi-dine residues in proteins (52, 53).

These differences, initially purely chemical, seemincreasingly to have a major impact on the responseof the immune system. The selectivity of the sites ofhaptenisation is directly involved in the selection ofthe peptide fragments that are presented by theAPCs to the T-cells, and thus in the selection of T-cell receptors. This selectivity also indirectly con-trols the level of haptenisation of the protein, orproteins. In recent years, the interest in radicalmechanisms as part of the hapten–protein bindingprocess has increased. This mechanism has neverbeen firmly established, but was postulated toexplain, for example, the allergenic potential ofeugenol versus iso-eugenol (54). More recently,studies have been published, which indicated thatradical reactions were important for haptens con-taining allylic hydroperoxide groups (55, 56).

Structure–Activity Relationships (SARs)

As outlined above, considerations of the chemicalproperties of known skin sensitisers, and compar-isons with non-sensitisers, have led to the conclu-


Figure 3: Principal electrophilic groups seen in contact allergy, with mechanisms of reactionand products

Group Name Reaction mechanism Product

R—CH2—X Alkyl halide Nucleophilic substitution Nu—CH2—RX = Cl, Br, I on a saturated centre

Aryl halide Nucleophilic substitution onan unsaturated centre

Aldehyde, R´ = H NucleophilicKetone, additionR´ = alkyl or aryl

Ester, R´ = OR Nucleophilic Amide, R´ = NHR substitution on an

unsaturated centre

Epoxide Nucleophilic substitution ona saturated centre

Lactone, X = O Nucleophilic Lactame, X = NH substitution on an

unsaturated centre

Aldehyde or ketone Nucleophilicα,β-unsaturated addition

p-quinone Nucleophilicaddition

o-quinone Nucleophilicaddition

Ni2+, Co2+, Cr(IV) Metal salts Coordinationbonds

NuHX

O

OX

X = F, Cl, Br, I

NO2

NO2

X

O

R = H, R, OR

R

O

O

O

O

Nu

OH

OH

Nu

OH

OH

Ni2+

LL

L L

NO2

NO2

Nu

R R´

O

R R´

O

O

R Nu

O

R Nu R´

OH

O

Nu

R

OH

Nu


sion that binding takes place by the action of theprotein as a nucleophile and the sensitiser as anelectrophile. One approach to estimating the skinsensitisation potential of a chemical is an “in silicoapproach” to look for electrophilic characteristics inthe structure of a molecule. This report is notintended to give an exhaustive overview of all theSAR and QSAR models developed for skin sensiti-sation, which have been reviewed elsewhere (28).The intention is to provide a short summary ofsome available models, and how hapten–protein

reactivity is considered in these theoreticalapproaches.

The general reactions such as those describedabove, are the basis for QSAR models developed topredict the skin sensitisation potentials of chemi-cals. In QSAR models, the ability of chemicals (hap-tens) to react with proteins to form covalently-linked conjugates is correlated with their skin sen-sitisation capability. For all such models, there is aneed to estimate the hapten–protein interactions.This is done either qualitatively, by evaluating thepresence or absence of a specific substructure in amolecule, or quantitatively, by using electronicdescriptors for estimating the potential reactivity ofa molecule.

A number of qualitative attempts have beenmade by using protein reactivity features and thepotential electrophilicities of molecules to deriverules for the identification of potential skin sensi-tisers. For example, a set of structural alerts basedon structural requirements for reactivity weredefined by Payne and Walsh (57). The alerts wereclassified as far as possible according to anticipatedreaction mechanisms, and, where applicable, wereincorporated into the expert system DEREK forWindows (58). Ashby et al. (59) and later, Roberts etal. (60, 61), suggested broad SARs for sensitisingchemicals, by grouping them into eight classesbased on their potential electrophilic properties,either directly or after metabolic activation. Gerneret al. (62) derived two sets of new structural alertsby using a regulatory database of sensitisers toestablish rules. Some SAR and structural alertshave been encoded into expert systems, e.g. TIMES-SS, Topkat, M-CASE, DEREK for Windows, andothers, some of which have been already reviewed(28, 63–65).

Various approaches have been taken to developQSAR models. For skin sensitisation, the availablepublished QSAR models fall into two main cate-gories; either they are mechanism-based (local mod-els) or they are derived empirically by usingstatistical approaches (global models). The quantifi-cation of hapten–protein interactions was first car-ried out by Roberts and Williams (27), whodeveloped a model known as the relative alkylationindex (RAI) model. It correlated the degree of car-rier alkylation (or the potential protein reactivity ofa chemical) with its sensitisation potential. The RAImodel approach has been used to evaluate a widerange of different datasets of skin sensitising chem-icals. It has been particularly successful for thedevelopment of QSAR models for small sets of struc-turally-similar chemicals, i.e. for producing localclass-specific models (60, 61, 66, 67). The alkylationpotential or the hapten–protein interaction poten-tial were estimated either by using measured reac-tivity data (reactivity to a model nucleophile, such asbutylamine) or by making use of Taft’s and Hammetcoefficients for some types of chemicals. The RAI

Figure 4: Principal nucleophilic residues inproteins

N

H

H

H

N

O

OH–CH2R =

O

N

H

–CH2

R =

N

N

H

R =

R = —(CH2)4—NH2 Lysine

Cysteine

Methionine

Histidine

Tryptophan

Tyrosine

R = —CH2—SH

R = —(CH2)2—S—Me

R


models can be defined as mechanism-based QSARmodels (or Quantitative Mechanistic Models[QMM]; 68). By grouping compounds into chemicalclasses and according to their reaction mechanisms,the RAI models can be further developed for struc-turally more-diverse sets of chemicals. Some prom-ising examples for using mechanistic applicabilitydomains for the prediction of toxicological endpointshave recently been proposed (69).

In contrast, the global models, which are oftenbased on large datasets involving diverse struc-tures, are mostly not mechanism based. They“encode” the hapten–protein interaction potentialby using one or a variety of descriptors which areoften difficult to interpret with respect to theunderlying biological hypothesis (70). The first

models were developed by Magee et al. (71), byusing a count of toxicophores to reflect reactivity.Cronin and Basketter (12) derived a global QSARmodel by using structural features associated withreactivity as descriptors, as well as Shannonindices and HOMO/LUMO differences. In 2005,Fedoro wicz et al. (72) found some topologicaldescriptors, geometry descriptors and topologicaldescriptors to be relevant for the development of askin sensitisation QSAR model. Topologicaldescriptors were also used by Estrada et al. (73) forthe development of a skin sensitisation classifica-tion model. All of these global models use statisti-cal techniques or a set of descriptors, but there isno attempt to rationalise the underlying sensitisa-tion mechanism.

Figure 5: Examples of sensitising molecules

Electrophilic centres are indicated by arrows.

O

�

�

�

�

�

O

alantolactone

O

HOOH

OO

linonene-1,2-oxide

O

(R)-carvone

� NO2

NO2

F

2,4-dinitrofluorobenzene

CH3

O

O

O

. HCl

N

N

H

�

propacetamol

21-dehydrohydrocortisone

O


In summary, hapten–protein interactions aredescribed either qualitatively or quantitatively byusing in silico models, and then, often combinedwith additional data, are used for the developmentof a QSAR model to predict the skin sensitisationpotentials of chemicals. The underlying reactionmechanisms are identical to those described in theprevious section.

Bioactivation or Activation ThroughInteraction with the Environment

Increased understanding of the importance of acti-vation through interaction with the environment toturn non-reactive compounds into skin sensitisershas made it important to distinguish pre-haptensfrom the pro-haptens by naming them differently.As outlined above, the activation of non-reactivemolecules into sensitisers can occur inside the skinby bioactivation via metabolism in the case of pro-haptens, or, in the case of pre-haptens, by chemicalactivation in the skin or outside the body prior toskin penetration, by interaction with the environ-ment (74) — with air oxidation (autoxidation) orphotoactivation being the most common mecha-nisms.

This distinction facilitates discussions on activa-tion, by emphasising the differences between thetwo types of compounds that need activation tobecome haptens. It is important to note that pre-hapten activation, in contrast to the bioactivation ofpro-haptens, could be prevented by precautionarymeasures in the handling and storage of the com-pounds. In the development of predictive test meth-ods, it will be important to include compoundsrepresenting both types of activation.

The bioactivation of pro-haptens

Numerous studies have shown that the human skinis able to metabolise endogenous and exogenouscompounds. The purpose of the metabolic activity isto detoxify these compounds, and to increase theirhydrophilicity, so that they are readily cleared viathe urine or bile. Both phase I and phase II enzymesare present in the skin. Cutaneous enzymes whichcatalyse phase I transformations include thecytochrome P450 (CYP) mixed-function oxidasesystem, alcohol and aldehyde dehydrogenases(ADHs and ALDHs), monoamine oxidases (MAOs),flavin-containing monooxygenases (FMOs), andhydrolytic enzymes. Acyltransferases, glutathioneS-transferases, UDP-glucuronosyltransferases andsulphotransferases are examples of phase IIenzymes, which have been shown to be present inhuman skin.

If the metabolites of phase I reactions are suffi-ciently polar, they may be readily eliminated at this

point. However, many phase I products are noteliminated rapidly and undergo subsequent phaseII biotransformations. Phase II metabolism consistsof conjugation reactions where substituents, eitherthose introduced by phase I metabolism or thosealready present in the compound, are conjugatedwith a hydrophilic endogenous derivative. Theamounts of cutaneous enzymes can be up to 1,000-fold lower than those found in the liver. However,when the enzyme activity per gramme of tissue isconsidered in relation to the large surface area ofthe skin (~2m2), it is clear that the skin is animportant metabolising organ (75).

The purpose of metabolism is detoxification, butthe same mechanisms can lead to the conversion ofinherently harmless compounds into reactive toxicspecies (76). This can be carried out by metabolicenzymes (e.g. CYPs) or can be non-enzymatic (i.e.via intramolecular rearrangements or intermolecu-lar reactions). The metabolites thus formed areusually electrophiles, but toxic free radical metabo-lites have also been suggested (77). Although not allcovalent modifications may be harmful, a largenumber of studies (e.g. 77, 78) suggest that theinadequate detoxification of reactive intermediatesleads to tissue necrosis, carcinogenicity, terato-genicity and immunotoxicity (e.g. contact allergy).

Few thorough mechanistic investigations on pro-haptens in contact allergy have been published upto now, e.g. the studies on the alpha,beta-unsatu-rated primary alcohol, cinnamic alcohol (Figure 6;79–81), on alkenes (Figure 7; 82–84), and on oximes(85). However, based on knowledge of xenobioticbioactivation reactions, clinical observations and/orin vivo and in vitro studies on sensitising capacityand chemical reactivity, numerous skin sensitisingpro-haptens can be recognised. The pro-haptensidentified include natural products, e.g. urushiols(86) and α-terpinene (83), dyes, e.g. p-phenylenedi-amine (87–89), flavours and fragrances, e.g. eugenol(90), drugs, e.g. sulphamethoxazole (91) and hydro-cortisone (92), and industrially-used chemicals, e.g.styrene (93) and ethylenediamine (94).

The air oxidation of pre-haptens

Most organic compounds can undergo autoxidation, areaction that can be defined as the insertion of oxy-gen into a C–H bond, forming a hydroperoxide(R–OOH). Autoxidation is a free radical chain reac-tion (Figure 8). The relative success of this slow reac-tion lies in the bi-radical character of the ground stateof oxygen. The oxygen acts as a fast and efficient rad-ical quencher, which results in the formation of a per-oxy radical (ROO·). The overall rate of the reaction istherefore determined by the ease of the hydrogenabstraction in the second step of the propagationreaction, resulting in the hydroperoxide (ROOH) anda new radical (R·) that feeds the radical chain reac-


Figure 6: A proposed mechanism for the bioactivation of cinnamic alcohol to cinnamic acid

A proposed mechanism for the bioactivation of cinnamic alcohol, firstly to cinnamic aldehyde by a CYP or alcoholdehydrogenase (ADH), followed by detoxification to cinnamic acid by aldehyde dehydrogenase (ALDH) or by ADHacting as a dismutase (79, 81–82).

O

H

O

OHOH

CYP or ADH

cinnamic alcohol

cinnamic acid

cinnamic aldehyde

ALDH or ADH

tion. This step is generally selective, and hydrogensin, for example, the benzylic, allylic and tertiary posi-tions, or next to heteroatoms, such as O and N, arepreferably abstracted, as this yields particularly sta-ble radicals.

Many terpenes are excellent targets for autoxida-tion, as they contain numerous easily abstractableallylic protons, a prerequisite for the autoxidationchain reaction. Thus, they are able to formhydroperoxides upon exposure to air, as demon-strated for linalool (95, 96; Figures 9 and 10). Thehydroperoxides are found to be the most potent sen-sitisers identified in the autoxidation mixtures.However, the hydroperoxides formed might not besufficiently stable, and may immediately degrade toless sensitising secondary oxidation products (e.g.alpha,beta-unsaturated carbonyl compounds andalcohols). Thus, in the terpenes studied to date, thesensitising capacity of the corresponding autoxi-dised mixtures could be correlated with the abilityto form stable hydroperoxides (Figure 11).

Colophony (rosin) obtained from coniferous treesis one of the most common causes of contact allergy.The major resin acids, abietic acid and dehydroabi-etic acid (Figure 12), are pre-haptens, and formpotent sensitisers upon exposure to the air (97, 98).The allylic tertiary hydroperoxide, 15-hydroperox-yabietic acid, a primary oxidation product from abi-etic acid, has been shown to be a major hapten incolophony (99).

Hydroperoxides are believed to form antigens viaa radical mechanism, starting with the homolyticcleavage of the O–O bond. The resulting alkoxy rad-ical may either bind directly with a protein or


Figure 7: A proposed mechanism for thebioactivation of a methylidene byCYP metabolism

See references 83–85 for further details.

O

prohapten

skinmetabolism

CYP

CYP

hapten

hapten

O

Figure 8: The general mechanism ofautoxidation

R–H + In· → R· + In–H

R· + O2 → R–OO·

R–OO· + R–H → R–OOH + R·

R–OO· + R· → R–OO–R

gInitiation

Propagation

Termination

Figure 9: Possible positions forhydroperoxide formation inlinalool

OH

Figure 10: Linalool and its identified oxidation products after autoxidation

OH

linalool

OH

OOH

OOH

OOH

OH

linalyl hydroperoxides

OH

O

OH

O

HO

O

HO

OH

OH

OH

OH

OH

linalyl alcohol

OH

linalyl aldehyde


rearrange to an epoxide, to form a hapten–carriercomplex. It has been argued that the hydroxy radi-cal (HO·) might act as a hapten, thus affording anunspecific allergic response. However, no suchunspecific responses were observed when cross-reactivity studies with hydroperoxides were per-formed in guinea-pigs (100).

The importance of developing methods able alsoto predict the sensitising capacities of haptensformed by autoxidation, has been demonstrated inclinical studies. Extensive studies in thousands ofpatients in various dermatology clinics in Europe,revealed that both autoxidised limonene andlinalool caused contact allergy to the same extentas the most common allergens in the standard trayfor the patch testing of contact allergens(101–104).

Photoactivation

Potentially, some organic molecules (pre-haptens)could also undergo activation by ultraviolet (UV)light. For this to occur, the absorption of UV energyby the sensitiser in the skin is usually required forboth induction and elicitation. For photoallergy, ithas been originally proposed that UV light causes aphotochemical reaction on the skin, not in the skin,that converts a photoallergen (pre-hapten) to a

stable contact allergen (105). An alternative mech-anism has been proposed, namely, that a photoal-lergen is activated by UV light to an unstableexcited molecule, which, under appropriate condi-tions, can interact with other molecules (e.g. pro-teins) normally found in the skin, to form anantigen or hapten (106). However, the irradiation ofa photoallergen in vitro, followed by repeated topi-cal application of the irradiated solution, is ineffec-tive in inducing photocontact sensitisation. Hence,the formation of a stable photoproduct from a pre-hapten outside the skin which can then act as apotential contact sensitiser, is a possible, but notvery likely, mechanism (107).

Degradation by microorganisms on the skin

Cutaneous microorganisms might also provide astimulus for allergic skin reactions. However, thishas been reported mainly in the context of atopiceczema, for which abnormal bacterial skin colonisa-tion is a characteristic feature (108). Micro -organisms do play a role in the pathogenesis ofatopic dermatitis, interacting with disease-suscepti-bility genes. Even though it is theoretically possible,no data have yet been published that demonstratethat skin microorganisms could degrade pre-hap-tens to form sensitising molecules.

Figure 11: Dose–response curves for linalool hydroperoxides, air-exposed linalool, linalylaldehyde, linalyl alcohol, and linalool, tested in the LLNA

Dose–response curves for linalool hydroperoxides ( ), linalool after a 45-week air-exposure ( ), linalool after a 10-weekair-exposure ( ), linalyl aldehyde ( ), linalyl alcohol ( ) and linalool ( ), tested in the LLNA. The horizontal, dottedline marks a stimulation index (SI) of 3, which is the cut-off limit for a compound to be considered a sensitiser (96, 97).

0

14

12

10

8

6

4

2

00.5

SI

1.0 1.5 2.0 2.5concentration (M)

3.0 3.5 4.0


Approaches for DeterminingChemical Reactivity TowardBiologically Relevant Nucleophiles

When setting up a methodology for measuring chem-ical reactivity in the context of in vitro tests for skinsensitisation, a number of choices have to be made,which, ultimately, will determine the applicability ofthe methodology on the one hand, and the level andsophistication of information gained on the otherhand. While in most cases, especially for direct-actingelectrophiles, a distinction between reactive or non-reactive molecules should be obvious from the struc-ture, this knowledge only allows for the simplequalitative classification of the chemical. As discussedbelow, qualitative app roaches are of value in identify-ing hazard, but they do not help to address the criti-cal need to understand allergenic potency and theultimate risk of skin sensitisation.

While there are a variety of ways of measuringreactivity, most assays monitor either the disap-pearance of a nucleophile or the formation of anadduct between the electrophile and the nucle-ophile. The means of measuring reactivity stronglyinfluences the choice of the appropriate analyticalmethod (see below). Another important choice isthe selection of an experimental nucleophile, whichcan range from a simple low molecular weight(LMW) organic compound (such as butylamine) upto a full-size, native protein (such as serum albu-min). Table 1 lists a number of relevant approachesand their key parameters, and some of theapproaches are discussed in greater detail below.

The detection of adduct formation versusdepletion of the nucleophile

As outlined in-depth above, the formation of a cova-lent adduct between a small electrophile and a pro-tein in the skin is assumed to be a hallmark of thesensitisation process. Therefore, it would appear log-ical that any in vitro assay on reactivity shoulddirectly monitor the formation of such adducts.Indeed, as can be seen in Table 1, the majority of thepublished studies did monitor adduct formationbetween selected skin sensitisers and model nucle-ophiles. This approach has the great advantage that,if adducts are found and their structure can beinferred from the analytical data, a mechanism ofadduct formation can be postulated. However, thechemical nature of adducts is often not predictable,and adducts with different test chemicals can varywidely in their physicochemical properties, thusrequiring specific modifications to the analyticalmethodology used for their detection. This maymake the standardisation of an alternative testbased on adduct formation rather difficult. In addi-tion, a negative result can either indicate that noadduct is formed or that the chosen method cannot

detect it, which then could lead to a false-negativeresult (i.e. it could reduce the sensitivity of themethod).

The contrary approach is to measure the loss ofthe target nucleophile in the presence of the testchemical. With this approach, no information isgained on the nature of the reaction taking placebetween the test chemical and the nucleophile.

Figure 12: Chemical structures of abieticacid, dehydroabietic acid, and15-hydroperoxyabietic acid

Abietic acid and dehydroabietic acid are the major resinacids in colophony. 15-hydroperoxyabietic acid is apotent sensitiser, formed from abietic acid byautoxidation (98).

COOH

COOH

OOH

COOH

Abietic acid

Dehydroabietic acid

15-hydroperoxyabietic acid


Tab

le 1

: A

rev

iew

of

sele

cted

pep

tid

e re

acti

vity

stu

die

s o

n c

on

tact

sen

siti

sers

Ad

du

ct v

ersu

sT

ype

ofP

epti

de

Ch

arac

teri

stic

sd

eple

tion

An

alyt

ical

P

aram

eter

s ch

emic

als

seq

uen

ceof

pep

tid

em

easu

rem

ent

met

hod

of a

nal

ysis

test

edK

ey r

esu

lts

Ref

.

Glu

tath

ione

(G

SH)

Phy

siol

ogic

al p

epti

de,

Add

ucts

LC

–MS

Qua

litat

ive

(+/–

)14

sen

siti

sers

,A

dduc

t fo

r 13

/14

sens

itis

ers,

no

addu

ct

(110

)re

acti

ve C

ys a

nd f

ree

4 no

n-fo

r 4/

4 no

n-se

nsit

iser

s, r

athe

r st

rong

N

-ter

min

al N

H2

grou

pse

nsit

iser

sse

nsit

iser

s te

sted

Glu

tath

ione

(G

SH)

Phy

siol

ogic

al p

epti

de, r

eact

ive

Dep

leti

onSp

ectr

opho

to-

Qua

ntit

ativ

e,

17 s

ensi

tise

rs, 6

EC

50 <

3.5

4mM

(>

25

exce

ss o

f te

st

(26,

70)

Cys

and

fre

e N

-ter

min

al

met

ric

EC

50 v

alue

sno

n-se

nsit

iser

s,ch

emic

al o

ver

GSH

) as

cut

-off

cor

rect

ly

NH

2 gr

oup

1 do

ubtf

ul

iden

tifi

es 1

0 ou

t of

17

sens

itis

ers,

no

com

poun

dfa

lse

posi

tive

s

Glu

tath

ione

(G

SH)

Phy

siol

ogic

al p

epti

de,

Add

uct

form

atio

n L

C–M

SQ

ualit

ativ

e(+

/–)

7 di

ffer

ent

GSH

-add

ucts

wit

h ox

idat

ed c

ompo

unds

(8

4)re

acti

ve C

ys a

nd f

ree

afte

r ac

tiva

tion

al

kene

sfo

r al

l tes

t ch

emic

als

in p

rese

nce

of

N-t

erm

inal

NH

2 gr

oup

by m

icro

som

esm

etab

olic

sys

tem

onl

y, c

hem

ical

nat

ure

and

not

pres

ence

/abs

ence

of

GSH

-add

ucts

co

rrel

ates

wit

h se

nsit

isat

ion

pote

ntia

l

PH

CK

RM

Des

igne

d pe

ptid

e co

ntai

ning

A

dduc

tsM

S an

d N

MR

Qua

litat

ive

(+/–

)T

wo

benz

ochi

none

A

dduc

ts f

ound

wit

h bo

th c

hem

ical

s(1

11)

all k

ey n

ucle

ophi

les

deri

vati

ves

PH

CK

RM

Des

igne

d pe

ptid

e co

ntai

ning

A

dduc

tsL

C–M

SQ

ualit

ativ

e (+

/–)

(5R

)-5-

isop

rope

nyl-

Add

ucts

wit

h th

e ep

oxy-

met

abol

ites

,but

(8

3)al

l key

nuc

leop

hile

s2-

met

hyl-1

-no

t w

ith

the

pare

nt c

ompo

unds

met

hyle

ne-2

-cy

cloh

exen

e an

d tw

o ep

oxid

e m

etab

olit

es

VL

SPA

DK

TN

WG

HE

YR

MF

QIG

N-t

erm

inal

por

tion

of

glob

in

Add

ucts

NM

RQ

ualit

ativ

e (+

/–)

MC

I/M

IA

dduc

ts f

or M

CI

wit

h L

ys a

nd H

is(5

2–53

)ch

ain

wit

h C

ys-1

8 re

mov

ed,

all o

ther

nuc

leop

hilic

gro

ups

pres

ent

AcR

FA

AC

AA

, AcR

FA

AK

AA

, D

esig

ned

pept

ides

, eac

h D

eple

tion

HP

LC

–DA

DQ

uant

itat

ive,

%

38 c

hem

ical

s w

ith

22 o

f 27

sen

siti

sers

, but

onl

y 1

out

of 1

1 (2

4)A

cRF

AA

HA

A, a

nd G

SHco

ntai

ning

1 k

ey n

ucle

ophi

lede

plet

ion

at f

ixed

tes

t kn

own

skin

no

n-se

nsit

iser

s, g

ave

sign

ific

ant

depl

etio

nch

emic

al a

nd p

epti

de

sens

itis

ing

dose

pote

ntia

l

AcR

FA

AC

AA

, AcR

FA

AK

AA

As

abov

eA

s ab

ove

As

abov

eA

s ab

ove

82 c

hem

ical

s w

ith

Cla

ssif

icat

ion

mod

el d

evel

oped

wit

h 89

%

(25)

know

n sk

in s

ensi

t-pr

edic

tion

acc

urac

yis

ing

pote

ntia

l

AcR

FA

AC

AA

, AcK

YD

CE

QR

, M

ost

sens

itiv

e pe

ptid

e fr

om

Dep

leti

on a

nd

HP

LC

–DA

D,

Qua

ntit

ativ

e,

67 c

hem

ical

s w

ith

29 o

f 44

sen

siti

sers

,but

onl

y 4

out

of 2

3 (1

12)

AcD

SRC

KD

Y, A

cKR

VC

EE

F,

abov

e st

udie

s; a

lter

nati

ve

addu

ct f

orm

atio

nL

C–M

S an

d (%

dep

leti

on),

kn

own

skin

no

n-se

nsit

iser

s,ga

ve s

igni

fica

nt d

eple

tion

AcN

KK

CD

LF

, AcQ

AN

CY

EE

pe

ptid

es d

eriv

ed f

rom

fl

uoro

met

ric

qual

itat

ive

sens

itis

ing

and

GSH

phys

iolo

gica

lly-r

eact

ive

dete

ctio

nfo

r ad

duct

spo

tent

ial

prot

eins


Whether a covalent adduct is indeed formed is notdirectly proven by this approach, as the test chemi-cal may also serve as a catalyst for the degradationof the nucleophile (for example, it may induce aredox-cycling process, oxidising the nucleophile).The importance of such unspecific reactions withproteins in the skin-sensitisation process is notestablished. On the other hand, the great advantageof this approach is the uniform measurementparameter of tests based on depletion of the testnucleophile. For any given test chemical, the sameparameter is measured, and the same analyticalmethodology is applied, so it is easily standardised,which permits comparisons to be made between testchemicals with widely different physicochemicalproperties. Given this significant advantage, meas-uring the depletion of a test nucleophile appears tobe a reasonable choice for the development of analternative assay for skin sensitisation. However, itshould be established whether, and under whichconditions, such a test result would really correlatewith real adduct formation, as the basic assumptionis that nucleophile depletion is indeed an indicatorof adduct formation.

Analytical detection methods

Whether specific adducts are sought, or only thedepletion of a standard nucleophile is to be deter-mined, will directly influence the choice of theappropriate analytical method. Adduct formationwith peptides is most often measured by LiquidChromatography–Mass Spectrometry (LC–MS)analysis, but in some cases, Nuclear MagneticResonance (NMR) and mass spectrometry withouta liquid chromatographic step have also been used.For adducts with proteins, a tryptic digest, followedby matrix-assisted laser desorption/ionisation–time-of-flight–mass spectrometry (MALDI–TOF–MS) analysis can indicate that a reaction hasoccurred, by observing the appearance of new sig-nals with mass shifts of modified peptides corre-sponding to hapten adducts. The exact localisationof the covalent modification can be determined bythe use of nano liquid chromatography MS/MS(nano-LC–MS/MS) strategies. These analyses per-mit the unambiguous identification of the exact siteof adduction on a large protein.

Various specific methods can be used to measurethe depletion of a nucleophile. HPLC–DAD (DiodeArray Detector) is a convenient, specific and accu-rate method, if the target nucleophile contains achromophore, e.g. from a Phe, Trp or Tyr residue.In addition, methods employing a molecular probewhich specifically reacts with the free nucleophileleft over at the end of the incubation period withthe test chemical, can be used. Probes for the fluo-rometric or UV-VIS spectrophotometric detectionof both free NH2 and SH groups are commercially

available. The big advantage of these methods istheir ease of use and the potential for a high-throughput set-up in microtitre plates, whichallows quick measurements of dose–responsecurves and kinetics (see below). The disadvantageof these methods is that the molecular probe itselfforms an adduct with the target nucleophile, and inthe case of only semi-stable adducts of the testchemicals with the nucleophile, a competitionbetween the molecular probe and the test chemicalcan lead to an underestimation of the true level ofadduct formation.

Finally, ELISA assays, which involve the reactionof a specific antibody with the target nucleophile,have been proposed (112), the concept being thatcovalent adducts between the epitope and the testchemical will prevent antibodies from binding,which again could lead to a high-throughput assayfor measuring the depletion of a nucleophile.

Parameters to describe reactivity

The simplest read-out of reactivity tests is of abinary nature, i.e. the chemical either does or doesnot react with the target nucleophile. As can beseen in the Table 1, most of the studies on adductformation have reported results of this qualitativenature. In principle, a quantitative measure couldalso be added to this approach, as LC–MS can givequantitative results. However, as the specificadduct normally is not available in a pure form,standard curves cannot be produced, which makesthe exact quantification of adducts with differenttest chemicals rather difficult.

A qualitative result is sufficient to determine thechemical reactions taking place and to investigatethe chemical nature of the adducts formed. It mayalso be enough for hazard identification, but in anideal setting, reactivity tests should add to hazardcharacterisation, and quantitative informationwould be very useful in this respect. This quantifi-cation is easier to obtain with depletion assays, andthe HPLC–UV, HPLC–MS and fluorometric/spec-trophotometric methods yield highly accurate andreproducible quantitative results.

Different parameters can then be chosen to quan-tify the reactivity. Each approach has its pros andcons in relation to time and resource requirements,and the level of information attained. The moststraightforward method is to express reactivity as %depletion of the nucleophile (i.e. incubate a fixedconcentration of test chemical with a fixed concen-tration of nucleophile, then determine the amountof free nucleophile remaining at a given time point).This approach makes the selection of the appropri-ate concentration of both test chemical and nucle-ophile most critical, as only with an optimal choiceof these concentrations can the relevant spectrumof differing reactivity be covered.


More-detailed results are obtained with concen-tration at a given time point–response curves, withvarying concentrations of a test chemical and afixed concentration of the target nucleophile, whileagain measuring depletion at a fixed incubationtime (113). Such concentration–response curves canyield RC50 values, which are algebraically relatedto the rate constant and independent of the experi-mental methodology. Even more-accurate quantifi-cation can be achieved by making kineticmeasurements at different time points to determinethe rate constant, k. Typically, in kinetic studies,one to several concentrations of a test chemical anda fixed concentration of the target nucleophile aretested, and measurement is performed at severaltime intervals. It is believed that rate constants areindependent of the method by which they are deter-mined. However, it is still a matter of debate as towhether these more elaborate and detailed kineticmeasurement methods are really needed to evalu-ate reactivity for predicting the skin sensitisationpotentials of chemicals.

Nucleophiles used in testing

A more critical feature of any reactivity method isthe choice of the target nucleophile. Nucleophilictargets in biological molecules are mostly the elec-tron-rich heteroatoms — sulphur, nitrogen, andoxygen. The biological nucleophiles associated withskin sensitisation, in particular, are thought to beproteins, which have many soft nucleophilic sitesthat include the thiol-group of cysteine, the primaryamino-groups of lysine and arginine, and the sec-ondary amino-group of histidine.

In deciding on a model nucleophile, a choice can bemade between LMW organic compounds, aminoacids, selected peptides, or proteins. Since the poten-tial target proteins in vivo are currently unknown,selecting a protein as the model nucleophile is bur-dened with undue ambiguity. Peptides, and especiallydesigner peptides, could provide a degree of biologicalrelevance that could be tailored to reflect a particularprotein microenvironment, but only with concomi-tant increase in complexity and cost.

Small molecular nucleophiles

Given that the nature and location of the carrierprotein(s) relevant to skin sensitisation are not cur-rently known, it cannot yet be decided which par-ticular model nucleophile, in which particularsolvent, would provide the most realistic model fora specific test approach. However, it can be arguedthat the choice of the model nucleophile for deter-mining reactivity is mainly dependent on how wellthe nucleophile performs in predicting sensitisationpotential. For example, Landsteiner and Jacobs (40)

used aniline in ethanol to discriminate betweenreactive and non-reactive halo-containing andnitro-containing aromatic compounds. This, inturn, enabled them to discriminate between sensi-tisers and non-sensitisers. In other examples, thereactivity of butylamine in various organic solventshas been used to obtain correlations with sensitisa-tion data for sultones (27), p-nitrobenzylhalides(114), and α-(halo-substituted methyl)-γγ-dimethyl-γ-butyrolactones (115, 116).

Peptides

By definition, the term “peptides” includes any-thing from simple dipeptides to polypeptides with100 amino acid residues. Peptides ranging fromdipeptides to polypeptides with 20 amino acidresidues have been used for reactivity assays (Table1). Test peptides can be designed to contain a singlenucleophilic residue. This permits the determina-tion of which nucleophilic species a given chemicalhas reacted with. However, the use of different testpeptides in parallel is necessary, if a spectrum ofnucleophiles is to be covered. Another approach isto use peptides designed to contain several relevantnucleophilic residues, in order to allow the testchemical to react with multiple nucleophiles withinthe same assay (82).

Test peptide sequences can be designed de novo,as is the case for the peptides specifically designedto either contain one specific nucleophile or severaldifferent nucleophiles (24, 82). The other approachis to use a physiological peptide, or at least asequence derived from a physiological protein.Thus, the simplest and most easily available testpeptide (which has also been most often used) is thetripeptide, glutathione (GSH). GSH is both themost abundant cellular peptide and the most abun-dant cellular thiol, and it is well known to reactboth spontaneously and enzymatically (catalysed byglutathione-S-transferases) with many electrophilicspecies. Other peptides have been derived fromphysiological proteins, such as the peptide derivedfrom globin, used by Alvarez-Sanchez et al. (52),and the peptides derived from different physiologi-cally reactive proteins, described by Natsch et al.(111). However, although these peptides have theirprimary structures in common with the originalproteins, their three-dimensional conformationswill be significantly different. Nevertheless, it wasshown that a highly reactive test peptide can beobtained by this approach.

Another study involved the use of a syntheticpeptide, PHCKRM, which contains common nucle-ophilic amino acids, to investigate the binding oftwo sensitising molecules, 1,4-benzoquinone and 4-t-butyl-1,2-benzoquinone (110, 117). The same pep-tide was used in a different study, to show that thetwo epoxide metabolites of (5R)-5-isopropenyl-2-


methyl-1-methylene-2-cyclohexane, but not theparent compound, bound to the Cys residue (82).Further examples of a variety of different peptidesused for studying reactivity of chemicals, but not inthe context of skin sensitisation, can be found in theliterature (118–121).

Proteins

Reactivity studies with model nucleophiles and pep-tides do not reveal the total spectrum of reactivityor any specificity in binding that may occur whenincubating a chemical with intact proteins contain-ing all the potential nucleophiles simultaneously.Owing to the complexity of the skin proteome (122)and the current lack of insight into immunogeni-cally-relevant protein targets in the skin, studies onthe reactivity of sensitising chemicals with proteinshave largely focused on the use of single model pro-teins. Elahi et al. (81) suggest that a moderate sen-sitising chemical (cinnamaldehyde) extensivelybinds to practically all proteins when incubatedwith skin homogenates. However, to date, there isno indication that such extensive binding occurs inintact human skin in vivo. Hopkins et al. (123) haveindicated that the protein targets of skin and respi-ratory sensitisers may be different. In fact, fromstudies which investigated specific target proteinsof electrophilic chemicals in complex mixtures(unrelated to skin sensitisation), it can be con-cluded that there may indeed be target proteins (orgroups of proteins) modified specifically by chemi-cal allergens (124, 125). It is possible that similartargeting may occur in the skin during sensitisa-tion. However, investigating these phenomena stillrepresents a considerable technical challenge.

In choosing a potential protein for studying thereactivities of chemicals, a number of factors haveto be considered, including the level of character-isation, relevance and relative abundance in skin,purity, solubility and price. For these reasons,human serum albumin (HSA) is by far the mostwidely-used model protein in such studies, butchemical modifications of other proteins, includingstructural and signalling proteins and even otheralbumins, have been described (50, 53, 126). Byemploying modern analytical techniques, whichoften include a combination of 1H and 13C NMR,MALDI–TOF–MS and nano electrospray tandemmass spectrometry (nano-ES–MS/MS), it is possibleto determine the exact localisations of covalentmodifications of sensitising chemicals. The modifi-cations observed in such studies are reproducible,and are highly selective for the few nucleophilesthat are presumably residing in specific microenvi-ronments particularly conducive to reactivity. Forexample, an investigation of the reactivities of threechemical sensitisers, MCI (53), 2,4-dinitro-1-chlorobenzene (DNCB) and phenyl salicylate (126),

with HSA, found that all three sensitisers cova-lently modified only selected nucleophilic residuesof HSA, in the case of MCI via more than one chem-ical mechanism. The data revealed remarkableinsights about the specificity of chemical modifica-tions of proteins, as each chemical modified a dif-ferent portfolio of residues, and only a few HSAresidues appeared to be commonly targeted by allthree chemicals. Whilst such insights are useful forgaining a detailed understanding of the nature andselectivity for specific proteins and protein nucle-ophiles, they are labour-intensive and their techni-cal complexity prevents them from being useful ingeneral screening.

The current status of the emerging methodologies

Since reactivity is one key step in the induction ofskin sensitisation, investigators have been inter-ested in pursuing whether measuring a chemical’sreactivity could be used to develop a quantitativepeptide-based reactivity assay that would be usefulfor screening a chemical’s skin sensitisationpotency, as defined in the LLNA. A few selectedreactivity methods which are being developed forsuch a purpose, are described below.

Kato et al. (109) proposed a GSH-peptide bindingassay involving the use of LC–MS and MALDI–MSanalyses as a screening method for skin sensitisa-tion. The assay is based on the detection of GSHconjugates following a 1-hour incubation withchemicals solubilised in acetone or water at 37°C. Atotal of 14 skin sensitisers and 4 non-sensitiserswere evaluated. This procedure identified conju-gates for 13 of the 14 sensitisers, and no conjugateswere identified for the 4 non-sensitisers (sensitivity93%, specificity 100%, total concordance 94%). Theassay did not identify conjugate(s) of the pre-haptenp-phenylenediamine, which is a strong sensitiser inin vivo tests. Mass spectrometric analysis withMALDI–TOF–MS revealed that compounds such asDNCB bound to the sulphydryl group of GSH. It isimportant to note that GSH also has an α-aminogroup that could be available for reactivity withskin sensitisers. The authors stated that MS has anumber of advantages over other analytical meth-ods. Specifically, it permits the characterisation ofthe conjugates and provides insight into reactionmechanisms, as well as providing better sensitivity.

Schultz and his associates have examined thevalue of using a non-enzymatic GSH reactivityassay as a potential non-animal approach to skinsensitisation testing (26, 113). The assay is a simpleand rapid spectrophotometric-based concentra-tion–response assay for non-enzymatic chemicalreactivity, following the incubation of GSH with thechemical. GSH contains the thiol group of cysteine,which is a primary nucleophile. Various concentra-


tions of the test compound, dissolved in dimethylsulphoxide, are mixed with a fixed concentration ofGSH for 2 hours at room temperature. The chro-mophore, 5,5´-dithio-bis(nitrobenzoic acid) wasused to measure the % free thiol as GSH. Definitiveconcentration–response experiments were run todetermine the 2-hour RC50 GSH values. The testconcentrations were adjusted so that at least threepartial responses were obtained, with at least oneon each side of the 50% effect concentration. TheRC50 thiol values were determined by ProbitAnalysis of Statistical Analysis System softwarefrom nominal chemical concentrations, wherechemical concentration was the independent vari-able, and absorbance normalised to the control levelwas the dependent variable. The RC50 thiol valueswere reported in mM units. Thus, a thiol concen-tration-specific/time-specific toxicant concentrationreflected a specific situation in a dynamic process.Aptula et al. (26) reported data for 24 substancesthat showed that chemicals which are considered toact as direct-acting electrophiles are most reactivein the GSH thiol assay, whereas chemicals consid-ered to be non- and weak electrophiles were eithernon- or weakly reactive with GSH. Acrolein is runas a positive control in each experiment, and datafrom multiple experiments demonstrated the excel-lent reproducibility of the assay. Current RC50thiol reactivity data are now available on over 200compounds, which represent various mechanisticdomains (e.g. Michael acceptors and SNAr elec-trophiles). Finally, it is very important to note thatthis rapid and easy assay provides a means to obtainrate constant data for potential use in in silico mod-els.

By using GSH and nucleophile-containing syn-thetic peptides, Gerberick et al. (24, 25) have evalu-ated the utility of these peptides to screen for skinsensitisation potential, by measuring peptide deple-tion following incubation with allergens and non-allergens. The GSH was incubated for 15 minuteswith the test chemical at a chemical:peptide ratio of1:100. Following incubation, the GSH was deriva-tised, and depletion of the GSH was monitored byHPLC. For the synthetic heptapeptides that con-tain either cysteine or lysine, the ratio of peptide tochemical used was 1:10 and 1:50. Following a 24-hour reaction period for the two synthetic peptides,the samples were analysed by HPLC with UV detec-tion, to monitor the depletion of the peptide follow-ing reaction.

The initial results with 38 chemicals representingallergens of different potencies (weak to extreme)and non-sensitisers, indicated a strong correlationbetween allergen potency and depletion of the non-reacted peptides containing cysteine or lysine (24).In general, moderate, strong and extreme sensitis-ers showed moderate to high reactivity, while weakand non-sensitisers showed minimal to low reactiv-ity. The analysis was expanded by evaluating a total

of 82 chemicals for their ability to react with GSHand the two synthetic peptides containing cysteineand lysine (25). The chemicals represented in thedataset comprised weak (n = 15), moderate (n =19), strong and extreme sensitisers (n = 18), as wellas non-sensitising materials (n = 30), as based onpotency categorisation criteria that have beendeveloped by the European Centre for Ecotox -icology and Toxicology of Chemicals (ECETOC).The peptide reactivity data were compared withexisting LLNA data, by using recursive partitioningmethodology to build a classification tree thatallowed a ranking of reactivity as minimal, low,moderate or high. In general, non-allergens andweak allergens demonstrated minimal to low pep-tide reactivity, whereas moderate to extremelypotent allergens displayed moderate to high peptidereactivity. Although a number of different modelswere developed that incorporated all or some of thenucleophile containing peptides, a model basedsolely on the cysteine peptide at 1:10 and the lysinepeptide at 1:50 was proposed as a method for cate-gorising reactivity by using this approach of meas-uring peptide depletion at a fixed time andconcentration. The cysteine and lysine peptides rep-resent softer to harder model nucleophiles, whichshould help in detecting skin sensitisers (elec-trophiles) which have different reaction mecha-nisms. To evaluate the approach for hazardidentification purposes, Cooper statistical analysiswas used. Classifying minimal reactivity as non-sensitisers, and low, moderate and high reactivityas sensitisers, it was determined that a model basedon cysteine and lysine gave a prediction accuracy of89%. Recently, an inter-laboratory study wasdesigned to evaluate the peptide reactivity of 15chemicals. Measurements of peptide reactivity weretaken for two synthetic peptides containing either acysteine or a lysine residue. The majority of thechemicals showed similar peptide depletion values(data not shown). In addition, the data wereanalysed by using the classification tree (recursivepartitioning) methodology described above, whichshowed that each of the three participating labora-tories classified reactivity similarly for > 90% of thechemicals evaluated.

Natsch et al. (111) evaluated the utility of a pep-tide reactivity assay for identifying fragrance aller-gens in vitro. The approach of measuring peptidedepletion was similar to the one described byGerberick et al. (24). However, additional peptidesand other approaches were also investigated. Theresults demonstrated that use of the cysteine pep-tide, as described in the Gerberick et al. paper (24),gave good results for strong and moderate fra-grance allergens, but was less robust for weakerallergens. It is important to note that some of thefragrance allergens that were not detected, areknown to be pro-haptens and would be likely torequire bioactivation if they were to be detected in


this assay system. Importantly, very few of the non-sensitisers did not react with the cysteine contain-ing peptide. The use of the fluorescent detectionmethod for the free thiol group was also demon-strated, and has the potential to provide higherthroughput analysis. Overall, there was very goodconcordance for most of the chemicals. The authorsalso evaluated dose–response curves, instead ofusing a fixed concentration. Finally, alternative testpeptides were derived from actual proteinsequences of highly reactive proteins, then evalu-ated by using the same approach, e.g. Cys420 ofhuman Coronin 1C; Cys138 in Cofilin (124).Overall, all of the different heptapeptides evaluated(n = 7) gave similar results and ranking of the com-pounds, although the Cor 1 peptide was slightlymore reactive than the other peptides.

To maximise the amount of qualitative and quan-titative information about the reactivity of a chem-ical with protein nucleophiles, a novel testingstrategy for the assessment of peptide reactivity hasrecently been proposed (127). This approach coversthe broad spectrum of nucleophiles: six peptideswith the generic sequence (AcFAAXAA, where Xrepresents a nucleophilic residue C, K, Y, H, R orW), and an additional peptide (H2N-FAAAAA) rep-resenting the N-terminus, have been sythesised forthis purpose; thus, all the test peptides contain onlyone nucleophilic residue. This approach is in theearly stages of development, and the testing strat-egy envisaged consists of a hazard identificationand a hazard characterisation. Following a 24-hourincubation with a chemical at 37°C, hazard identifi-cation is based on the observed depletion of any ofthe nucleophilic peptides, in addition to the detec-tion of a covalent peptide-adduct or oxidative dam-age, by LC–MS. The hazard characterisationincludes the determination of the chemical reactionmechanism(s), reaction specificities, and reactionkinetics measured by quantitative and qualitativeLC–MS/MS. These analyses are generating a richdataset for integration with the data from otherpredictive assays in novel weight-of-evidence-basedrisk assessment approaches for skin sensitisation.However, this assay can provide for only a lowerthroughput of chemicals, as compared with theother assays described above.

There are a few important points for considerationwhen interpreting data generated by using thesepeptide reactivity assays, whether they are designedto look at peptide depletion or at conjugate forma-tion. First, the assays have been developed primarilyfor evaluating sensitisers that are capable of directlyreacting with nucleophiles containing peptides orproteins, and only few pro- or pre-haptens have beenevaluated to date. Second, there is no consensus overwhich nucleophiles are best for use in reactivityscreening assays. Third, there can be interferencewith some electrophiles, when spectrophometric-based methods are used. Finally, the non-physiologi-

cal conditions used for some of these approaches,that seem more compatible with aqueously solublematerials, may prove challenging for the analysis ofhighly lipophilic compounds. Of course, this is a chal-lenge for any aqueous-based in vitro approach whichis under development, whether it is chemically orbiologically based. However, the overall results of theinvestigations described above reveal that measure-ment of peptide reactivity has potential utility as ascreening approach for skin sensitisation testing,and thereby for reducing the current reliance on ani-mal-based test methods.

As a test chemical may be a pre-hapten or a pro-hapten, it will be critical to incorporate methodsthat allow for either spontaneous air-oxidation(simulating hapten formation by product ageing) ormetabolic activation (simulating the enzymaticactivation processes of the pro-hapten in the skin).Air oxidation can be performed easily, and the for-mation of haptens by spontaneous oxidations(mainly peroxides, but also epoxides) has beenrepeatedly reported (95, 96, 128). Thus, it may beenvisaged that an air oxidation step would beincluded for any test chemical. Biological activationis much more difficult to perform in vitro. Themost-detailed experience comes from pharmaceuti-cal research, where target drugs are routinely incu-bated with liver microsomes and GSH in order todetect potential GSH-adducts of the metaboliseddrugs (129). That a similar approach is also valid forskin sensitisers was shown by Bergström et al. (83).Pro-haptens were incubated with GSH and withliver microsomes. The pro-haptens were metab -olised, and GSH adducts of the resulting metabo-lites could be detected by LC–MS. However,metabolites and GSH-adducts were also formedwith non-sensitising derivatives, and it was thenature of the adducts (resulting from epoxides),rather then the mere presence of an adduct, whichwas indicative of the sensitisation potential. Thus,significant expert knowledge is needed to judgethese data, and they demonstrate the risk involvedin using a highly-active metabolic compartment.Many compounds could be activated to GSH reac-tive species, even if they are not efficientlymetabolised to haptens in the skin. In an extensionof this work (83), a CYP enzyme cocktail closelymimicking the CYP composition found in the skin,was designed, and it was shown that sensitisingepoxides are also formed with this more skin-likeenzyme cocktail. Whether this cocktail predomi-nantly forms reactive species from known skin-sen-sitisers/pro-haptens, or whether it also activatesmany proven non-sensitisers, is a key question forfurther research.

A more-pragmatic approach to the enzymaticactivation of pro-haptens was chosen by Troutmanet al. (130), which involves the use of peroxide andperoxidase as a bioactivation system. Specifically, aquantitative enzyme-based in vitro method involv-


ing horseradish peroxidase and hydrogen peroxide(HRP/P) has been developed for measuring the pep-tide reactivities of pro-hapten chemical sensitisers.The assay includes enzymatic (+HRP/P) and non-enzymatic (–HRP/P) reactivity determinations forthe test chemical (200µM) with a cysteine-basedsynthetic peptide (20µM). Sample analysis is per-formed by using a rapid quantitative HPLC–MS/MSdetection method that is simple and amenable tohigh-throughput screening. The peptide reactivityfor 27 pro-hapten and non-sensitising chemicalsover a broad range of skin sensitisation potencieshas been determined. Trends in peptide depletionwere highly reproducible, and were in good agree-ment with in vivo sensitisation data. In conjunctionwith other in vitro methods, this assay could be veryuseful for assisting in the assessment of a chemi-cal’s skin sensitisation potential by using a bioacti-vation system.

Summary and Conclusions

1. For a chemical to induce an immune response,it must associate in stable fashion with amacromolecule; in practice a protein or glyco-protein.

2. The acquisition of skin sensitisation is there-fore dependent on the formation of stable asso-ciations between the inducing chemicalallergen and protein. Such a chemical allergenis defined immunologically as being a hapten.

3. Therefore, skin haptens are either inherentlyprotein-reactive or can be activated in the skinto a protein-reactive species.

4. Chemicals can form associations with proteinvia various molecular mechanisms. These canvary from weak and reversible interactions(hydrophobic, dipolar, ionic), to those that arestrong and stable (covalent bond formation).

5. For the vast majority of haptens, it is covalentinteractions with protein that are required forthe formation of an immunogenic complexthat will facilitate the acquisition of skin sen-sitisation.

6. Metal allergens represent a special case, inthat, although strong associations with proteinare required for immunogenicity, these neednot be irreversible (coordination bonds).

7. It is not clear where and when immunogenicadducts are formed during the induction phaseof skin sensitisation.

8. Nor is it apparent which proteins within theskin are involved in the formation of immuno-genic adducts.

9. It is probable that, in practice, a variety of pro-tein adducts will be formed in the skin.However, not all of these will necessarily beimmunogenic, or will contribute to the induc-tion of a cutaneous immune response.

10. It must also be acknowledged that the forma-tion of protein adducts is a dynamic process,during which the creation of new adducts canoccur in tandem with the degradation of con-jugated proteins.

11. Although the formation of immunogenic cova-lent adducts in the skin is a very complicatedprocess, certain aspects can be modelled invitro.

12. Thus, it has proven possible to quantitativelymeasure the formation by reactive chemicalsof covalent bonds with model proteins and pep-tides.

13. It is not clear whether or not adducts gener-ated in vitro reflect conjugates that will beformed by the same chemicals in vivo.

14. Nevertheless, adduct formation with modelproteins and peptides in vitro shows real prom-ise as an alternative approach for the identifi-cation of skin sensitisation hazards.

15. For the identification in such systems of pro-haptens (that require conversion to protein-reactive species), some form of bioactivationmay be required — and some progress hasbeen made in this context. However, ourunderstanding of the processes involved isincomplete.

16. It is not yet clear in what ways adduct forma-tion with protein influences the intrinsic skinsensitisation potency of a chemical allergen.

17. Against this background, it is similarly unclearwhat specific aspects of adduct formation maybe relevant for the measurement of relativepotency (amino acid selectivity, reaction rate,stability of protein conjugates, etc.).

18. It is important to acknowledge that, althoughthere may be opportunities for to developing‘stand-alone’ tests for skin sensitisation haz-


ard identification based on chemical reactivity,it is also relevant to consider how peptide/pro-tein reactivity methods could contribute tointegrated testing strategies.

Recommendations

1. The current methods for the identification ofskin sensitisation hazards based on the meas-urement of protein/peptide reactivity showpromise, and are worthy of further invest-ment.

2. Specifically, standard protocols should now beidentified, and initial prediction models shouldbe defined, that will facilitate further evalua-tion and validation.

3. In addition to the methods identified in thisreport, there are other approaches based onthe measurement of protein/peptide reactivity,that also appear to have the potential to iden-tify skin sensitisation hazards. It is recom-mended that these approaches are alsoconsidered for exploitation as test methods forskin sensitisation.

4. There is a need for continued investment inthe development of methods and approachesthat will permit the appropriate activation ofpro-haptens and pre-haptens.

5. It is necessary that the further development,refinement and validation of test methodsbased on protein/peptide reactivity, is under-pinned by a continued investment in relevantareas of basic scientific research and techno-logical development.

6. One area of uncertainty that represents animportant research objective, is how peptide/protein reactivity measured in vitro relates toevents in the skin.

7. There is also a need to consider how methodsbased on the measurement of peptide/proteinreactivity can be aligned with otherapproaches, in order to provide integratedstrategies for skin sensitisation testing.

8. A major focus of future research should be theinvestigation of ways in which the inherent orpotential reactivities of chemical allergensimpact on relative skin sensitising potency.

9. In a practical context, it will be important toidentify the parameters that will permit themeasurement of protein/peptide reactivity, insupport of the assessment of potency.

10. A follow-up workshop will need to be convenedat some future date, to assess progress in thedevelopment of skin sensitisation tests basedon reactivity, and to encourage further invest-ment, if required, and as appropriate.

References

1. Anon. (1967). EEC Council Directive 67/548/EEC onthe approximation of the laws, regulations andadmin istrative provisions relating to the class i -fication, packaging and labelling of dangerous substances. Official Journal of the European Com -munities 196, 1–98.

2. Anon. (1999). Directive 1999/45/EC of the EuropeanParliament and of the Council of 31 May 1999 con-cerning the approximation of the laws, regulationand administrative provisions of the Member Statesrelating to the classification, packaging andlabelling of dangerous preparations. OfficialJournal of the European Communities L200, 1–68.

3. Anon. (2006). Regulation (EC) No 1907/2006 of theEuropean Parliament and of the Council of 18December 2006 concerning the Registration, Eval -uation, Authorisation and Restriction of Chemicals(REACH), establishing a European ChemicalsAgency, amending Directive 1999/45/EC and repeal-ing Council Regulation (EEC) No 793/93 andCommission Regulation (EC) No 1488/94 as well asCouncil Directive 76/769/EEC and Comm issionDirectives 91/155/EEC, 93/67/EEC, 93/105/EC and2000/21/EC. Official Journal of the European UnionL396, 1–849. Available at: http://eur-lex.europa.eu/LexUriServ/site/en/oj/2006/l_396/l_39620061230en00010849.pdf (Accessed 14.04.08).

4. Magnusson, B. & Kligman, A.M. (1970). AllergicContact Dermatitis in the Guinea Pig, 141pp.Spring field, IL, USA: Charles C. Thomas.

5. Buehler, E.V. (1965). Delayed contact hypersensi-tivity in the guinea pig. Archives of Dermatology 91,171–177.

6. Kimber, I. & Basketter, D.A. (1992). The murine locallymph node assay; collaborative studies and newdirections: A commentary. Food & Chemical Toxicol -ogy 30, 165–169.

7. Anon. (1999). The Murine Local Lymph NodeAssay: A Test Method for Assessing the AllergicContact Dermatitis Potential of Chemicals/Com -pounds. The Results of an Independent Peer ReviewEvaluation Coordinated by the Interagency Coordin -ating Committee on the Validation of AlternativeMethods (ICCVAM) and the National ToxicologyProgram Center for the Evaluation of AlternativeToxicological Methods (NICEATM). NIH Publica -tion 99-4494, 211pp. Research Triangle Park, NC,USA: National Toxicology Program. Available at:http://iccvam.niehs.nih.gov/docs/immunotox_docs/llna/llnarep.pdf (Accessed 14.04.08).

8. Balls, M. & Hellsten, E. (2000). Statement on thevalidity of the local lymph node assay for skin sen-sitisation testing. ATLA 28, 366–367.

9. Gerberick, G.F., Ryan, C.A., Kimber, I., Dearman,R.J., Lea, L.J. & Basketter, D.A. (2000). Locallymph node assay validation assessment for regula-tory purposes. American Journal of Contact Derm -atitis 11, 3–18.

10. Dean, J.H., Twerdok, L.E., Tice, R.R., Sailstad,


D.M., Hattan, D.G. & Stokes, W.S. (2001). ICCVAMevaluation of the murine local lymph node assay. II.Conclusions and recommendations of an independ-ent scientific peer review panel. Regulatory Tox -icology & Pharmacology 34, 258–273.

11. Wahlberg, J.E. & Boman, A. (1985). Guinea pigmaximization test. In Contact Allergy PredictiveTests in Guinea Pigs: Current Problems in Derm -atology (ed. H.I. Maibach & K.E. Anderson), pp.59–106. New York, NY, USA: S. Karger.

12. Cronin, M.T.D. & Basketter, D.A. (1994). Multi -variate QSAR analysis of a skin sensitisation data-base. SAR & QSAR in Environmental Research 2,159–179.

13. Basketter, D.A. & Gerberick, G.F. (1996). Inter -laboratory evaluation of the Buehler test. ContactDermatitis 35, 146–151.

14. Anon. (1992). Council Directive 92/32/EEC of 30April 1992 amending for the seventh time Directive67/548/EEC on the approximation of the laws, regu-lations and administrative provisions relating to theclassification, packaging and labelling of dangeroussubstances. Official Journal of the European Com -munities L154, 1–29.

15. Anon. (2001). Commission Directive 2001/59/EC of6 August 2001 adapting to technical progress for the28th time Council Directive 67/548/EEC on theapproximation of the laws, regulations and admin-istrative provisions relating to the classification,packaging and labelling of dangerous substances(Text with EEA relevance). Official Journal of theEuropean Communities L225, 1–333.

16. Basketter, D.A., Andersen, K.E., Lidén, C., vanLoveren, H., Boman, A., Kimber, I., Alanko, K. &Berggren, E. (2005). Evaluation of the skin sensitis-ing potency of chemicals using existing methods andconsiderations of relevance for elicitation. ContactDermatitis 52, 39–43.

17. Kimber, I., Basketter, D.A., Butler, M., Gamer, A.,Garrigue, J.L., Newsome, C., Steiling, W. & Vohr,H-W. (2003). Classification of allergens according topotency: proposals. Food & Chemical Toxicology 41,1799–1809.

18. WHO (1997). Criteria for classification of skin andairway sensitizing substances in the work and gen-eral environments. Doc. No. EUR/ICP/EHPM 05 0201(P), 111pp. Copenhagen, Denmark: World HealthOrganisation, Regional Office for Europe.

19. Basketter, D.A., Gerberick, F. & Kimber, I. (2007).The local lymph node assay and the assessment ofrelative potency: status of validation. ContactDermatitis 57, 70–75.

20. Felter, S.P., Ryan, C.A., Basketter, D.A., Gilmour,N.J. & Gerberick, G.F. (2003). Application of therisk assessment paradigm to the induction of aller-gic contact dermatitis. Regulatory Toxicology &Pharmacology 37, 1–10.

21. Api, A.M., Basketter, D.A., Cadby, P.A., Cano, M-F.,Ellis, G., Gerberick, G.F., Griem, P., McNamee,P.M., Ryan, C.A. & Safford, B. (2008). Dermal sen-sitisation quantitative risk assessment (QRA) forfragrance ingredients. Regulatory Toxicology &Pharm acology, in press.

22. Jowsey, I., Basketter, D.A., Westmoreland, C. &Kimber, I. (2006). A future approach to measuringrelative skin sensitising potency. Journal of AppliedToxicology 26, 341–350.

23. Gerberick, G.F., Ryan, C.A., Kern, P.S., Schlatter, H.,Dearman, R.J., Kimber, I., Patlewicz, G. & Basketter,

D.A. (2005). Compilation of historical local lymphnode assay data for the evaluation of skin sensitisa-tion alternatives. Dermatitis 16, 157–202.

24. Gerberick, G.F., Vassallo, J.D., Bailey, R.E., Chaney,J.G., Morrall, S.W. & Lepoittevin, J.P. (2004).Development of a peptide reactivity assay for screen-ing contact allergens. Toxicological Sciences 81,332–343.

25. Gerberick, G.F., Vassallo, J.D., Foertsch, L.M., Price,B.B., Chaney, J.G. & Lepoittevin, J.P. (2007). Quant -ification of chemical peptide reactivity for screeningcontact allergens: A classification tree modelapproach. Toxicological Sciences 97, 417–427.

26. Aptula, A.O., Patlewicz, G., Roberts, D.W. & Schultz,T.W. (2006). Non-enzymatic glutathione reactivityand in vitro toxicity: a non-animal approach to skinsensitisation. Toxicology in Vitro 20, 239–247.

27. Roberts, D.W. & Williams, D.L. (1982). The deriva-tion of quantitative correlations between skin sensi-tisation and physico–chemical parameters foralkylating agents and their application to experi-mental data for sultones. Journal of TheoreticalBiology 99, 807–825.

28. Patelwicz, G., Aptula, A.O., Uriarte, E., Roberts,D.W., Kern, P.S., Gerberick, G.F., Kimber, I.,Dearman, R.J., Ryan, C.A. & Basketter, D.A. (2007).An evaluation of selected global (Q)SARs/expert sys-tems for the prediction of skin sensitisation potential.SAR & QSAR in Environmental Research 18,515–541.

29. Kimber, I., Basketter, D.A., Gerberick, G.F. & Dear -man, R.J. (2002). Allergic contact dermatitis.International Immunopharmacology 2, 201–211.

30. Dearman, R.J. & Kimber, I. (2003). Factors influenc-ing the induction phase of skin sensitisation.American Journal of Contact Dermatitis 14, 188–194.

31. Kissenpfennig, A. & Malissen, B. (2006). Langerhanscells — revisiting the paradigm using geneticallyengineered mice. Trends in Immunology 27, 132–139.

32. Kimber, I., Cumberbatch, M., Dearman, R.J.,Bhushan, M. & Griffiths, C.E.M. (2000). Cytokinesand chemokines in the initiation and regulation ofepidermal Langerhans cell mobilization. BritishJournal of Dermatology 142, 401–412.

33. Kimber, I. & Cumberbatch, M. (1992). Dendriticcells and cutaneous immune responses to chemicalallergens. Toxicology & Applied Pharmacology 117,137–146.

34. Kimber, I., Dearman, R.J., Cumberbatch, M. & Huby,R.J.D. (1998). Langerhans cells and chemical allergy.Current Opinion in Immunology 10, 614–619.

35. Cumberbatch, M., Dearman, R.J., Antonopoulos, C.,Groves, R.W. & Kimber, I. (2001). Interleukin (IL)-18 induces Langerhans cell migration by a tumournecrosis factor-α and IL-1β-dependent mechanism.Immunology 102, 323–330.

36. Wang, B., Feliciani, C., Howell, B.G., Freed, I., Cai,Q., Watanabe, H. & Sauder, D. (2002). Contributionof Langerhans cell-derived IL-18 to contact hyper-sensitivity. Journal of Immunology 168, 3303–3308.

37. Ryan, C.A., Gerberick, G.F., Gildea, L.A., Hulette,B.C., Betts, C.J., Cumberbatch, M., Dearman, R.J.& Kimber, I. (2005). Interactions of contact aller-gens with dendritic cells: opportunities and chal-lenges for the development of novel approaches tohazard assessment. Toxicological Sciences 88, 4–11.

38. Griffiths, C.E.M., Dearman, R.J., Cumberbatch, M.& Kimber, I. (2005). Cytokines and Langerhans cellmobilisation in mouse and man. Cytokine 32, 67–70.


39. Kimber, I. & Dearman, R.J. (2002). Allergic contactdermatitis: the cellular effectors. Contact Derm -atitis 46, 1–5.

40. Landsteiner, K. & Jacobs, J. (1936). Studies on thesensitisation of animals with simple chemical com-pounds. Journal of Experimental Medicine 64,625–639.

41. Dupuis, G. & Benezra, C. (1982). Allergic ContactDermatitis to Simple Chemicals: A MolecularApproach, 200pp. New York, NY, USA & Basel,Switzerland: Marcel Dekker.

42. Darley, M.O., Post, W. & Munter, R.L. (1977).Induction of cell-mediated immunity to chemicallymodified antigens in guinea pigs. Journal ofImmunology 118, 963–970.

43. Frierson, M.R., Klopman, G. & Rosenkranz, H.S.(1986). Structure–activity relationships amongmutagens and carcinogens: A review. Environ -mental Mutagenesis 8, 283–327.

44. Guengerich, F.P. & Liebler, D.C. (1985). Enzymaticactivation of chemicals. CRC Critical Reviews inToxicology 14, 259–307.

45. Means, G.E. & Feeney, R.E. (1971). Chemical Mod -if ication of Proteins, 254pp. San Francisco, CA,USA: Holden Day.

46. Romagnoli, P., Labahrdt, A.M. & Sinigaglia, F.(1991). Selective interaction of nickel with an MHCbound peptide. The EMBO Journal 10, 1303–1306.

47. Lepoittevin, J-P. & Benezra, C. (1992). 13C-enrichedmethyl alkanesulfonates: New lipophilic methylat-ing agents for the identification of nucleophilicamino acids of proteins by NMR. TetrahedronLetters 33, 3875–3878.

48. Franot, C., Benezra, C. & Lepoittevin, J-P. (1993).Synthesis and interaction studies of 13C labeled lac-tone derivatives with a model protein using 13C NMR.Bioorganic & Medicinal Chemistry 1, 389–397.

49. Meschkat, E., Barratt, M.D. & Lepoittevin, J-P.(2001). Studies of the chemical selectivity of hapten,reactivity, and skin sensitisation potency. 1: Syn -thesis and studies on the reactivity toward modelnucleophiles of the 13C-labeled skin sensitizers hex-1-ene- and hexane-1,3-sultones. Chemical Researchin Toxicology 14, 110–117.

50. Meschkat, E., Barratt, M.D. & Lepoittevin, J-P.(2001). Studies of the chemical selectivity of hapten,reactivity, and skin sensitisation potency. 2: NMRstudies of the covalent binding of the 13C-labeledskin sensitizers 2-[13C]- and 3-[13C]hex-1-ene- and3-[13C]hexane-1,3-sultones to human serum albu-min. Chemical Research in Toxicology 14, 118–126.

51. Alvarez-Sanchez, R., Basketter, D.A., Pease, C. &Lepoittevin, J-P. (2003). Studies of chemical selec-tivity of hapten, reactivity and skin sensitisationpotency. 3: Synthesis and studies on the reactivitytowards model nucleophiles of the 13C-labeled skinsensitizers, 5-chloro-2-methylisothiazol-3-one (MCI)and 2-methylisothiazol-3-one (MI). Chemical Res -earch in Toxicology 16, 627–636.

52. Alvarez-Sanchez, R., Basketter, D.A., Pease, C. &Lepoittevin, J-P. (2004). Covalent binding of the13C-labeled skin sensitizers 5-chloro-2-methylisoth-iazol-3-one (MCI) and 2-methylisothiazol-3-one(MI) to a model peptide and glutathione. BioorganicMedicinal Chemistry Letters 14, 365–368.

53. Alvarez-Sanchez, R., Divkovic, M., Basketter, D.A.,Pease, C., Panico, M., Dell, A., Morris, H. &Lepoittevin, J-P. (2004). Effect of glutathione onthe covalent binding of the 13C-labelled skin sensi-

tizer 5-chloro-2-methylisothiazol-3-one (MCI) tohuman serum albumin: identification of adducts byNMR, MALDI–MS and nano-ES–MS/MS. ChemicalResearch in Toxicology 17, 1280–1288.

54. Barratt, M.D. & Basketter, D.A. (1992). Possibleorigin of the skin sensitization potential of eugenoland related compounds. Contact Dermatitis 27,98–104.

55. Giménez-Arnau, E., Haberkorn, L., Grossi, L. &Lepoittevin, J-P. (2002). Identification of alkyl rad-icals derived from an allergenic cyclic tertiary allylichydroperoxide by combined use of radical trappingand ESR studies. Tetrahedron 58, 5535–5545.

56. Bézard, M., Gimenez-Arnau, E., Meurer, B., Grossi,L. & Lepoittevin, J.P. (2005). Identification by rad-ical trapping and EPR studies of carbon-centeredradicals derived from linalyl hydroperoxide, astrong skin sensitizer: a possible route for proteinmodifications. Bioorganic Medicinal Chemistry 13,3977–3986.

57. Payne, M.P. & Walsh, P.T. (1994). Structure–activ-ity relationships for skin sensitisation potential:development of structural alerts for use in knowl-edge based toxicity prediction systems. Journal ofChemical Information & Computer Sciences 34,154–161.

58. Barratt, M.D., Basketter, D.A., Chamberlain, M.,Payne, M.P., Admans, G.D. & Langowski, J.J.(1994). An expert system rulebase for identifyingcontact allergens. Toxicology in Vitro 8, 1053–1060.

59. Ashby, J., Basketter, D.A., Paton, D. & Kimber, I.(1995). Structure–activity relationships in skin sen-sitisation using the murine local lymph node assay.Toxicology 103, 177–194.

60. Roberts, D.W. & Basketter, D.A. (1997). Furtherevaluation of the quantitative structure–activityrelationship for skin-sensitizing alkyl transferagents. Contact Dermatitis 37, 107–112.

61. Roberts, D.W., York, M. & Basketter, D.A. (1999).Structure–activity relationships in the murine locallymph node assay for skin sensitisation: α,β-dike-tones. Contact Dermatitis 41, 14–17.

62. Gerner, I., Barratt, M.D., Zinke, S., Schlegel, K. &Schlede, E. (2004). Development and prevalidationof a list of structure–activity relationship rules to beused in expert systems for prediction of the skinsensitizing properties of chemicals. ATLA 32,487–509.

63. Karol, M.H., Graham, C. & Barratt, M.D. (1999).Structure–activity relationship studies of hypersen-sitivity. Toxicology 7, 59–77.

64. Bashir, S.J. & Maibach, H.I. (1999). Quantitativestructure analysis relationships in the prediction ofskin sensitisation potential. In Biochemical Modu -lation of Skin Reactions: Transdermals, Topicals,Cosmetics (ed. A.F. Kydonieus & J.J. Wille), pp.61–64. New York, NY, USA: CRC Press.

65. Rodford, R., Patlewicz, G., Walker, J.D. & Payne,M.P. (2003). Quantitative structure–activity rela-tionships for predicting skin and respiratory sensi-tisation. Environmental Toxicology & Chemistry22, 1855–1861.

66. Roberts, D.W. (1987). Structure–activity relation-ships for skin sensitisation potential of diacrylatesand dimethacrylates. Contact Dermatitis 17, 281–289.

67. Roberts, D.W. & Benezra, C. (1993). Quantitativestructure–activity relationships for skin sensitisa-tion potential of urushiol analogues. ContactDermatitis 29, 78–83.


68. Aptula, A.O., Patlewicz, G. & Roberts, D.W. (2005).Skin sensitisation: Reaction mechanistic applicabil-ity domains for structure–activity relationships.Chemical Research in Toxicology 18, 1420–1426.

69. Aptula, A.O. & Roberts, D.W. (2006). Mechanisticapplicability domains for nonanimal-based predic-tion of toxicological end points: General principlesand application to reactive toxicity. ChemicalResearch in Toxicology 19, 1097–1105.

70. Li, Y., Tseng, Y.J., Pan, D., Liu, J., Kern, P.S., Ger -berick, G.F. & Hopfinger, A.J. (2007). 4D-finger-print categorical QSAR models for skinsensitisation based on the classification of locallymph node assay measures. Chemical Research inToxicology 20, 114–128.

71. Magee, P.S., Hostynek, J.J. & Maibach, M.I. (1994).A classification model for allergic contact dermati-tis. Quantitative Structure–Activity Relationships13, 22–33.

72. Fedorowicz, A., Singh, H., Soderholm, S. & Dem -chuk, E. (2005). Structure–activity models for con-tact sensitization. Chemical Research in Toxicology18, 954–969.

73. Estrada, E., Patlewicz, G., Chamberlain, M.,Basketter, D. & Larbey, S. (2003). Computer-aidedknowledge generation for understanding skin sensi-tisation mechanisms: The TOPS-MODE approach.Chemical Research in Toxicology 16, 1226–1235.

74. Lepoittevin, J-P. (2006). Metabolism versus chemi-cal transformation or pro- versus prehaptens?Contact Dermatitis 54, 73–74.

75. Smith, C.K. & Hotchkiss, S.A.M. (2001). AllergicContact Dermatitis: Chemical and Metabolic Mech -anisms, 310pp. London, UK: Taylor and Francis.

76. Mabic, S., Castagnoli, K. & Castagnoli, N. (1999).Oxidative metabolic bioactivation of xenobiotics. InHandbook of Drug Metabolism (ed. T.F. Woolf), pp.49–79. New York, NY, USA: Marcel Dekker.

77. Kalgutkar, A.S., Gardner, I., Obach, R.S., Shaffer,C.L., Callegari, E., Henne, K.R., Mutlib, A.E.,Dalvie, D.K., Lee, J.S., Nakai, Y., O’Donnell, J.P.,Boer, J. & Harriman, S.P. (2005). A comprehensivelisting of bioactivation pathways of organic func-tional groups. Current Drug Metabolism 6, 161–225.

78. Guengerich, F.P. (2005). Principles of covalentbinding of reactive metabolites and examples ofactivation of bis-electrophiles by conjugation. Arch -ives of Biochemistry & Biophysics 433, 369–378.

79. Smith, C.K., Moore, C.A., Elahi, E.N., Smart, A.T.& Hotchkiss, S.A. (2000). Human skin absorptionand metabolism of the contact allergens, cinnamicaldehyde, and cinnamic alcohol. Toxicology &Applied Pharmacology 168, 189–199.

80. Cheung, C., Hotchkiss, S.A. & Pease, C.K. (2003).Cinnamic compound metabolism in human skin andthe role metabolism may play in determining rela-tive sensitisation potency. Journal of Dermat ol -ogical Science 31, 9–19.

81. Elahi, E.N., Wright, Z., Hinselwood, D., Hotchkiss,S.A., Basketter, D.A. & Pease, C.K. (2004). Proteinbinding and metabolism influence the relative skinsensitisation potential of cinnamic compounds.Chemical Research in Toxicology 17, 301–310.

82. Nilsson, A.M., Bergstrom, M.A., Luthman, K.,Nilsson, J.L. & Karlberg, A.T. (2005). A conjugateddiene identified as a prohapten: contact allergenicactivity and chemical reactivity of proposed epoxidemetabolites. Chemical Research in Toxicology 18,308–316.

83. Bergstrom, M.A., Luthman, K., Nilsson, J.L. &Karlberg, A.T. (2006). Conjugated dienes as prohap-tens in contact allergy: in vivo and in vitro studiesof structure–activity relationships, sensitizingcapacity, and metabolic activation. ChemicalResearch in Toxicology 19, 760–769.

84. Bergstrom, M.A., Ott, H., Carlsson, A., Neis, M.,Zwadlo-Klarwasser, G., Jonsson, C.A., Merk, H.F.,Karlberg, A.T. & Baron, J.M. (2007). A skin-likecytochrome P450 cocktail activates prohaptens tocontact allergenic metabolites. The Journal ofInvestigative Dermatology 127, 1145–1153.

85. Bergstrom, M.A., Luthman, K. & Karlberg, A.T.(2007). Metabolic epoxidation of an alpha,beta-unsaturated oxime generates sensitizers of extremepotency. Are nitroso intermediates responsible?Chemical Research in Toxicology 20, 927–936.

86. Goetz, G., Meschkat, E. & Lepoittevin, J.P. (1999).Synthesis of a water-soluble analog of 6-methyl-3-N-alkyl catechol labeled with carbon 13: NMRapproach to the reactivity of poison ivy/oak sensitiz-ers toward proteins. Bioorganic Medicinal Chem -istry Letters 9, 1141–1146.

87. White, J.M., Kullavanijaya, P., Duangdeeden, I.,Zazz eroni, R., Gilmour, N.J., Basketter, D.A. &McFadden, J.P. (2006). p-Phenylenediamine allergy:the role of Bandrowski’s base. Clinical & Experi -mental Allergy 36, 1289–1293.

88. Basketter, D.A. & Goodwin, B.F. (1988). Invest -igation of the prohapten concept. Cross reactionsbetween 1,4-substituted benzene derivatives in theguinea pig. Contact Dermatitis 19, 248–253.

89. Coulter, E.M., Farrell, J., Mathews, K.L., Maggs, J.L.,Pease, C.K., Lockley, D.J., Basketter, D.A., Park, B.K.& Naisbitt, D.J. (2007). Activation of human den-dritic cells by p-phenylenediamine. Journal ofPharmacology & Experimental Therapeutics 320,885–892.

90. Bertrand, F., Basketter, D.A., Roberts, D.W. & Lep -oittevin, J.P. (1997). Skin sensitisation to eugenoland isoeugenol in mice: possible metabolic path-ways involving ortho-quinone and quinone methideintermediates. Chemical Research in Toxicology 10,335–343.

91. Naisbitt, D.J. (2004). Drug hypersensitivity reac-tions in skin: understanding mechanisms and thedevelopment of diagnostic and predictive tests.Toxicology 194, 179–196.

92. Isaksson, M., Bruze, M., Goossens, A. & Lepoitt -evin, J-P. (2000). Patch-testing with serial dilutionsof tixocortol pivalate and potential cross-reactivesubstances. Acta Dermato Venereologica 80, 33–38.

93. Sjöborg, S., Fregert, S. & Trulsson, L. (1984).Contact allergy to styrene and related chemicals.Contact Dermatitis 10, 94–96.

94. Kimber, I., Hilton, J., Dearman, R.J., Gerberick,G.F., Ryan, C.A., Basketter, D.A., Lea, L., House,R.V., Ladics, G.S., Loveless, S.E. & Hastings, K.L.(1998). Assessment of the skin sensitisation poten-tial of topical medicaments using the local lymphnode assay: an interlaboratory evaluation. Journalof Toxicology & Environmental Health 53, 563–579.

95. Sköld, M., Börje, A., Matura, M. & Karlberg, A.T.(2002). Studies on the autoxidation and sensitizingcapacity of the fragrance chemical linalool, identify-ing a linalool hydroperoxide. Contact Dermatitis 46,267–272.

96. Sköld, M., Börje, A., Harambasic, E. & Karlberg,A.T. (2004). Contact allergens formed on air expo-


sure of linalool. Identification and quantification ofprimary and secondary oxidation products and theeffect on skin sensitisation. Chemical Research inToxicology 17, 1697–1705.

97. Karlberg, A.T. (1988). Contact allergy to colophony.Chemical identifications of allergens, sensitisationexperiments and clinical experiences. Acta DermatoVenereologica 139, 1–43.

98. Karlberg, A.T. (2000). Colophony. In Handbook ofOccupational Dermatology (ed. L. Kanerva, P.Elsner, J.E. Wahlberg & H.I. Maibach), pp.509–516. Heidelberg, Germany: Springer-Verlag.

99. Karlberg, A.T., Bohlinder, K., Boman, A., Hacksell,U., Hermansson, J., Jacobsson, S. & Nilsson, J.L.(1988). Identification of 15-hydroperoxyabietic acidas a contact allergen in Portuguese colophony.Journal of Pharmacy & Pharmacology 40, 42–47.

100. Brared-Christensson, J., Matura, M., Backtorp, C.,Börje, A., Nilsson, J.L.G. & Karlberg, A.T. (2006).Hydroperoxides form specific antigens in contactallergy. Contact Dermatitis 55, 230–237.

101. Karlberg, A.T. & Dooms-Goossens, A. (1997).Contact allergy to oxidized d-limonene among der-matitis patients. Contact Dermatitis 36, 201–206.

102. Matura, M., Goossens, A., Bordalo, O., Garcia-Bravo,B., Magnusson, K., Wrangsjo, K. & Karlberg, A.T.(2002). Oxidized citrus oil (R-limonene): a frequentskin sensitizer in Europe. Journal of the AmericanAcademy of Dermatology 47, 709–714.

103. Matura, M., Sköld, M., Börje, A., Andersen, K.E.,Bruze, M., Frosch, P., Goossens, A., Johansen, J.D.,Svedman, C., White, I.R. & Karlberg, A.T. (2005).Selected oxidized fragrance terpenes are commoncontact allergens. Contact Dermatitis 52, 320–328.

104. Matura, M., Sköld, M., Börje, A., Andersen, K.E.,Bruze, M., Frosch, P., Goossens, A., Johansen, J.D.,Svedman, C., White, I.R. & Karlberg, A.T. (2006).Not only oxidized R-(+)- but also S-(–)-limonene isa common cause of contact allergy in dermatitispatients in Europe. Contact Dermatitis 55, 274–279.

105. Willis, I. & Kligman, A.M. (1968). The mechanismof photoallergic contact dermatitis. Journal ofInvestigative Dermatology 51, 378.

106. Harber, L.C. & Bickers, D.R. (1989). Photosen sitivityDisease: Principles of Diagnosis and Treat ment,442pp. Philadelphia, PA, USA: W.B. Saunders.

107. Schidt, R.J. & Kingston, T. (1984). Testing withmusk ambrette and congeners in a case of photo-sensitivity dermatitis and actinic reticuloid syn-drome (PD/AR). Photodermatology 1, 195–198.

108. Baker, B.S. (2005). The role of microorganisms inatopic dermatitis. Clinical & Experimental Immun -ology Review 144, 1–9.

109. Kato, H., Okamoto, M., Yamashita, K., Nakamura,Y., Fukumori, Y., Nakai, K. & Kaneko, H. (2003).Peptide-binding assessment using mass spectrome-try as a new screening method for skin sensitisa-tion. Journal of Toxicological Sciences 28, 19–24.

110. Ahlfors, S.R., Sterner, O. & Hansson, C. (2003).Reactivity of contact allergenic haptens to aminoacid residues in a model carrier peptide, and char-acterization of formed peptide–hapten adducts.Skin Pharmacology & Applied Skin Physiology 16,59–68.

111. Natsch, A., Gfeller, H., Rothaupt, M. & Ellis, G.(2007). Utility and limitations of a peptide reactiv-ity assay to predict fragrance allergens in vitro.Toxicology in Vitro 21, 1220–1226.

112. Aeby, P., Griem, P., Gerberick, F. & Goebel, C.

(2007). Development and preliminary evaluation ofa monoclonal antibody based assay for detectingprotein reactive chemicals. The Toxicologist 96,248.

113. Schultz, T.W., Yarbrough, J.W. & Johnson, E.L.(2005). Structure–activity relationships for reactiv-ity of carbonyl-containing compounds with glu-tathione. SAR & QSAR in Environmental Research16, 313–322.

114. Roberts, D.W., Goodwin, B.F., Williams, D.L.,Jones, K., Johnson, A.W. & Alderson, J.C. (1983).Correlations between skin sensitisation potentialand chemical reactivity for p-nitrobenzyl com-pounds. Food & Chemical Toxicology 21, 811–813.

115. Franot, C., Roberts, D.W., Smith, R.G., Basketter,D.A., Benezra, C. & Lepoittevin, J-P. (1994).Structure–activity relationships for contact aller-genic potential of gamma,gamma-dimethyl-gamma-butyrolactone derivatives. 1: Synthesis andelectro philic reactivity studies of alpha-(omega-sub-stituted-alkyl)-gamma,gamma-dimethyl-gamma-butyrolactones and correlation of skin sensitisationpotential and cross-sensitisation patterns withstructure. Chemical Research in Toxicology 7,297–306.

116. Franot, C., Roberts, D.W., Basketter, D.A. &Benezra, C. (1994). Structure–activity relationshipsfor contact allergenic potential of gamma,gamma-dimethyl-gamma-butyrolactone derivatives. 2:Quantitative structure–skin sensitisation relation-ships for alpha-substituted-alpha-methyl-gamma,gamma-dimethyl-gamma-butyrolactones. ChemicalResearch in Toxicology 7, 307–312.

117. Hanson, C., Ahlfors, S. & Sterner, O. (1999). React -ivities of contact allergenic quinones towards amodel peptide and identification of the adducts.Journal of Investigative Dermatology 113, 320–320.

118. Tracey, B.M. & Shuker, D.E. (1997). Charact er iza -tion of azo coupling adducts of benzenediazoniumions with aromatic amino acids in peptides and pro-teins. Chemical Research in Toxicology 10,1378–1386.

119. Conduah Birt, J.E., Shuker, D.E. & Farmer, P.B.(1998). Stable acetaldehyde–protein adducts as bio-markers of alcohol exposure. Chemical Research inToxicology 11, 136–142.

120. Fabrizi, L., Taylor, G.W., Canas, B., Boobis, A.R. &Edwards, R.J. (2003). Adduction of the chloroformmetabolite phosgene to lysine residues of humanhistone H2B. Chemical Research in Toxicology 16,266–275.

121. Baker, A., Zidek, L., Wiesler, D., Chmelik, J., Pagel,M. & Novotny, M.V. (1998). Reaction of N-acetyl-glycyllysine methyl ester with 2-alkenals: an alter-native model for covalent modification of proteins.Chemical Research in Toxicology 11, 730–740.

122. Celis, J.E., Ostergaard, M., Jensen, N.A., Gromova,I., Rasmussen, H.H. & Gromov, P. (1998). Humanand mouse proteomic databases: novel resources inthe protein universe. FEBS Letters 430, 64–72.

123. Hopkins, J.E., Naisbitt, D.J., Kitteringham, N.R.,Dearman, R.J., Kimber, I. & Park, B.K. (2005).Selective haptenation of cellular or extracellularprotein by chemical allergens: association withcytokine polarization. Chemical Research in Toxic -ology 18, 375–381.

124. Dennehy, M.K., Richards, K.A., Wernke, G.R., Shyr,Y. & Liebler, D.C. (2006). Cytosolic and nuclear pro-tein targets of thiol-reactive electrophiles. Chemical


Research in Toxicology 19, 20–29.125. Shin, N.Y., Liu, Q., Stamer, S.L. & Liebler, D.C.

(2007). Protein targets of reactive electrophiles inhuman liver microsomes. Chemical Research inToxicology 20, 859–867.

126. Aleksic, M., Pease, C.K., Basketter, D.A., Panico,M., Morris, H.R. & Dell, A. (2007). Investigatingprotein haptenation mechanisms of skin sensitisersusing human serum albumin as a model protein.Toxicology in Vitro 21, 723–733.

127. Jacquoilleot, S., Thain, E., Zazzeroni, R. & Aleksic,M. (2007). Covalent modification of model peptides:Peptide reactivity testing strategies. Abstract 578 inProceedings of the 55th ASMS Conference on MassSpectrometry and Allied Topics. Santa Fe, NM,USA: American Society for Mass Spectrometry.

128. Hagvall, L., Backtorp, C., Svensson, S., Nyman, G.,Borje, A. & Karlberg, A.T. (2007). Fragrance com-pound geraniol forms contact allergens on air expo-sure. Identification and quantification of oxidationproducts and effect on skin sensitization. ChemicalResearch in Toxicology 20, 807–814.

129. Zheng, J., Ma, L., Xin, B., Olah, T., Humphreys,W.G. & Zhu, M. (2007). Screening and identificationof GSH-trapped reactive metabolites using hybridtriple quadruple linear ion trap mass spectrometry.Chemical Research in Toxicology, 20, 757–766.

130. Troutman, J., Gerberick, F., Foertsch, L., Quijano,M., Dobson, R. & Lepoittevin, J-P. (2007). Methodoptimization of a peroxidase peptide reactivity assayfor screening the skin sensitisation potential of pro-haptens. The Toxicologist 96, 248.


chemical reactivity measurement and the predictive ... · of non-animal test development ......

Documents