Skin Sensitization QMM for HRIPT NOEL Data: Aldehyde Schiff-Base

May 11, 2017 - Not all parts of the Schiff base domain are modeled with one equation. Particularly, predicting aromatic aldehydes and ketones appears ...
0 downloads 0 Views 892KB Size
Article pubs.acs.org/crt

Skin Sensitization QMM for HRIPT NOEL Data: Aldehyde Schiff-Base Domain David W. Roberts,*,† Terry W. Schultz,‡ and Anne Marie Api§ †

School of Pharmacy and Biomolecular Sciences, Liverpool John Moores University, Liverpool L3 3AF, United Kingdom College of Veterinary Medicine, The University of Tennessee, 2407 River Drive, Knoxville, Tennessee 37996, United States § Research Institute for Fragrance Materials, Inc., 50 Tice Boulevard, Woodcliff Lake, New Jersey 07677, United States ‡

ABSTRACT: The general chemistry principles underlying skin sensitization for Schiff base (SB) electrophiles may be used to develop a quantitative mechanistic model (QMM), based on reactivity supplemented with a hydrophobicity parameter for some but not all structures within the SB reaction domain. For aliphatic Schiff base electrophiles, the log of the no observed effect level (NOEL) values (pNOEL) from the human repeated insult patch test (HRIPT) can be calculated by the reactivity parameter summation of sigma star values (Σσ*) and a hydrophobicity parameter (logP). Specifically, the QMM, pNOEL = 2.34(±0.33) Σσ* + 0.19(±0.07) logP − 2.62(±0.22), n = 19, R2 = 0.77, R2(adj) = 0.74, s = 0.20, F = 27, was developed. Not all parts of the Schiff base domain are modeled with one equation. Particularly, predicting aromatic aldehydes and ketones appears to require a separate equation. Interestingly, the same physical organic chemical properties originally applied to modeling the local lymph node assay potency of Schiff base electrophiles apply to human potency as represented by the HRIPT.



As recently noted,20 nonanimal approaches to assessing skin sensitization include artificial neural networks, Bayesian networks, and decision trees. Quantitative mechanistic models (QMM) for predicting skin sensitization potential/potency21−24 are based on the relative alkylation index (RAI) concept of Roberts and Williams.25 The RAI concept is based on the principle that the degree of sensitization produced at induction and the magnitude of the sensitization response at challenge depend on the degree of covalent binding to carrier protein occurring at induction and challenge. Following this principle, chemicals are grouped by domain, and then a statistical (usually linear) relationship between (for example) the rate constant for reactivity and molar potency (e.g., LLNA pEC3) is derived. The Schiff base electrophile reaction mechanistic domain, consisting mainly of aldehydes, is well represented in fragrance chemical inventories, both natural and synthetic. Many of these chemicals are to some extent allergenic, and their sensitization potency in the LLNA has been modeled21 by a combination of a calculated reactivity parameter, Σσ*, and a hydrophobicity parameter, 1-octanol/water partition coefficient (logP). Σσ* is the sum of the Taft substituent constants, σ*, for the two groups bonded to the reactive carbonyl group. Taft σ* values are the aliphatic counterparts of the Hammett σ values used for

INTRODUCTION

Traditionally, the ability of chemicals to cause skin sensitization has been assessed, for risk assessment purposes, by guinea pig tests and, more recently, by the murine local lymph node assay (LLNA).1−6 Results from the human repeated insult patch test (HRIPT)7 can also be used for this purpose. The LLNA gives results in the form of a quantitative potency index, the EC3 value8−14 (concentration giving a 3-fold increase in lymph node activity), and has proved a useful source of data for modeling work. For the past two decades, there have been substantial initiatives aimed at developing nonanimal approaches to assess skin sensitization potential and potency.15 These nonanimal approaches are consistent with the 3Rs of reduction, refinement, and replacement.16 The most promising approach for quantitative modeling of skin sensitization is the application of chemical reaction mechanistic domains.17 This approach is based on the concept that different quantitative relationships exist between skin sensitization potency and chemical properties for each reaction mechanism. The advantage with the reaction mechanism approach is that the OECD skin sensitization adverse outcome pathway,18,19 whereby a reactive substance (parent compound or metabolite) undergoes the molecular initiating event (i.e., covalent interaction with skin proteins), initiates a cascade of other events, leading to skin sensitization, and underlies all mechanistic domains of skin sensitization. © 2017 American Chemical Society

Received: February 27, 2017 Published: May 11, 2017 1309

DOI: 10.1021/acs.chemrestox.7b00050 Chem. Res. Toxicol. 2017, 30, 1309−1316

Article

Chemical Research in Toxicology Table 1. Aliphatic Aldehydes Investigated and Their NOEL Valuesa

a

chemicals

CAS

NOELobs [μg/cm2]

pNOELobs

pNOELcalc

α-methyl-phenylacetaldehyde 1,2,3,4,5,6,7,8-octahydro-8,8-dimethyl-2-naphthaldehyde phenylacetaldehyde citral cuminyl acetaldehyde bourgeonal p-methylhydrocinnamic aldehyde p-isobutyl-α-methyl hydrocinnamaldehyde hydroxycitronellal lilial landolal (lyral) methoxy dicyclopentadiene carboxaldehyde triplal 2-methyl-3-(p-methoxyphenyl)propanal cyclamen aldehyde heptanal, 6-methoxy-2,6-dimethyl3-phenylbutanal citronellal isocyclocitral α-methyl-1,3-benzodioxole-5-propionaldehyde

93−53−8 68991−97−9 122−78−1 5392−40−5 7775−00−0 18127−01−0 5406−12−2 6658−48−6 107−75−5 80−54−6 31906−04−4 86803−90−9 68039−49−6 5462−06−6 103−95−7 62439−41−2 16251−77−7 106−23−0 1335−66−6 1205−17−0

388 551 591 775 (1400)b 1102 1181 1379 2362 2953 (5000)b 3750 (4125)b 4000 5000 5905 5905 5905 5905 5906 7086 7087 11811 (4016)b

−0.4611 −0.4572 −0.6917 −0.7069 −0.7959 −0.7928 −0.9687 −1.0630 −1.2341 −1.2638 −1.2792 −1.4105 −1.6307 −1.5203 −1.4918 −1.5349 −1.6005 −1.6620 −1.6680 −1.7885

−0.6950 −1.3518 −0.5034 −0.5823 −1.1140 −1.1184 −1.2337 −1.2283 −1.4126 −1.3100 −1.2163 −1.6412 −1.5836 −1.5121 −1.3056 −1.7432 −1.4267 −1.4126 −1.5418 −1.5260

Figures in the pNOELcalc column are calculated from regression eq 3. bValues from ref 44. QMM analysis. Aromatic aldehydes, for reasons that are not unequivocally clear, are usually significantly less potent (in some cases nonsensitizing) in the LLNA than would be expected from SAR considerations.21 Although there are insufficient aromatic aldehyde human potency data for QMM analysis, we can still use these data to compare the observed human potency with potency predicted from the aliphatic aldehyde QMM. Figure 1 and Figure 2 show the structures of the chemicals investigated in this analysis (aliphatic and aromatic aldehydes, respectively). The HRIPT (NOEL) data are compiled from two references.27,28 It is worth noting that the NOEL values taken from human experimentation are often best estimates derived from very limited data and are not necessarily the highest possible value; that is, they are a no-effect level rather than a level at which sensitization is produced.27 Bearing this in mind, where the two sources of data disagree, we used the higher of the two values for the regression analysis described below.27,28 The logP values were calculated as follows. First BioByte ClogP v1.6 software (http://www.biobyte.com) was used to generate initial logP values and outputs detailing the in silico calculations. These in silico calculations were checked manually and adjusted in cases where appropriate position dependent branch factors (PDBF) had not been applied. The in silico method uses branching factors of −0.13 and −0.22 (applied where the branch occurs at a carbon atom bonded to the aldehyde group), and these were replaced, as in the previous QMM exercise for Schiff base electrophiles,21 by the position dependent branching factor.21,31 In the present data set, all branches are methyl, for which the position dependent branching factor is −0.43.31 In some cases, methyl branches appeared to have been ignored by the in silico method. Table 2 gives the in silco logP values, the corrrections applied, and the corrected logP values. Taft σ* values were taken from the compilation of Perrin et al. or calculated using the methods presented by the same authors.32 The Σσ* values are shown in Table 2. Modeling. Following the procedure of Roberts et al.,21 a QMM for aliphatic SB aldehydes was generated by multiple linear regression analysis using MINITAB software version 17.0 (MINITAB Inc., State College, PA) with a combination of a calculated reactivity parameter (Σσ*) and a hydrophobicity parameter (ClogPcorr). For purposes of QMM analysis, the NOEL values were converted to a molar logarithmic parameter pNOEL = log (MW/NOEL), where MW is

aromatic systems and model the electronic effects that the substituent groups produce at the reaction center.21 The QMM derived, based on LLNA data for 11 aliphatic aldehydes, 1 α-ketoester and 4 α,β-diketones is pEC3 = 1.12(± 0.07)Σσ * + 0.42(± 0.04)logP − 0.62( ±0.13) n = 16, R2 = 0.952, R2(adj) = 0.945, s = 0.12, F = 129.6 (1)

The need for the logP parameter in this QMM suggests that in cutaneo the formation of SB adducts occurs in a hydrophobic medium such as a membrane. For this reactivity domain, logP models relative bioavailability in a lipid environment where the molecular initiating event (binding to protein) occurs. Roberts and Aptula26 define bioavailability as “the proportion of sensitizer applied to the outer skin which reaches the epidermis within the time scale of the skin sensitization assay (or other relevant exposure duration) and thereby becoming available to contribute to skin sensitization”. HRIPT data are available for a number of chemicals.27,28 These data are used, in addition to risk assessment purposes, to assess the predictive performance of animal tests and alternative methods.29,30 However, routine testing of new chemicals on humans, even in the HRIPT, is not a viable option. In an effort to reduce the need for clinical testing of chemicals, domain-based QMM building based on HRIPT derived no observed effect level (NOEL) values was undertaken. Specifically, we investigated if the same physical organic chemistry principles initially applied to modeling LLNA potency of SB electrophiles21 could be applied to human potency (HRIPT-derived NOEL values) with the same performance.



MATERIALS AND METHODS

Data Set. Table 1 shows the NOEL data for 20 aliphatic SB aldehydes. We were only able to find NOEL data for 4 aromatic aldehydes (Table 3), which we consider an insufficient number for 1310

DOI: 10.1021/acs.chemrestox.7b00050 Chem. Res. Toxicol. 2017, 30, 1309−1316

3.26 4.23 3.87 1.58 2.36 2.32 3.83 1.86

107−75−5 80−54−6 31906−04−4 86803−90−9 68039−49−6 5462−06−6 103−95−7 62439−41−2 16251−77−7 106−23−0 1335−66−6 1205−17−0

3-phenylbutanal citronellal isocyclocitral α-methyl-1,3-benzodioxole-5-propionaldehyde

1311

replace 1 FBr by 1 PDBF(Me) and add 1 more PDBF(Me) replace 1 FgBr and 1 FBr by 2 PDBF(Me) and add one more PDBF(Me) replace 1 FBr by 1 PDBF(Me) replace 2 FBr by 2 PDBF(Me) replace 3 FBr by 3 PDBF(Me) add one more PDBF(Me)

c

adjustments replace 1 FgBr by 1 PDBF(Me) replace 3 FBrd by 3 PDBF(Me) N/A replace 2 FBr by 2 PDBF(Me) replace 1 FBr by 1 PDBF(Me) replace 2 FBr by 2 PDBF(Me) N/A replace 1 FBr by 1 PDBF(Me) and add one more PDBF(Me) replace 2 FBr by 2 PDBF(Me) replace 2 FBr by 2 PDBF(Me) and add 1 more PDBF replace 1 FBr by 1 PDBF(Me) N/A replace 2 FBr by 2 PDBF(Me) add one more PDBF(Me)

b

1.97 2.66 1.98 1.94

3.10 0.92

2.66 3.20 3.57 1.58 1.76 1.89

1.88 2.98 1.78 2.35 3.00 3.10 2.37 3.63

logPcorr

0.35 0.30 0.30 0.31

0.31 0.30

0.30 0.30 0.31 0.29 0.30 0.32

0.67 0.30 0.76 0.68 0.40 0.39 0.40 0.30

Σσ*

a The ClogPcalc column shows the logP values calculated in silico. These were modified as outlined in the adjustments column to give the logPcorr values that were used, together with the Σσ* values, to derive eq 3. bFgBr = −0.22. cPDBF(Me) = −0.43. dFBr = −0.13. eThe log P values are for the unsaturated aldehydes resulting from the elimination of water, which is assumed to occur in vivo.21,45,46

2.27 3.26 2.88 2.37

2.09 3.79 1.78 2.95 3.30 3.70 2.37 4.36

93−53−8 68991−97−9 122−78−1 5392−40−5 7775−00−0 18127−01−0 5406−12−2 6658−48−6

α-methyl-phenylacetaldehyde 1,2,3,4,5,6,7,8-octahydro-8,8-dimethyl-2-naphthaldehyde phenylacetaldehyde citral cuminyl acetaldehyde bourgeonal p-methylhydrocinnamic aldehyde p-isobutyl-α-methyl hydrocinnamaldehyde (in the ClogP calculation the Me-branch alpha to the CO group is not allocated a branch factor) hydroxycitronellale lilial landolal (lyral)e methoxy dicyclopentadiene carboxaldehyde triplal 2-methyl-3-(p-methoxyphenyl)propanal (in the ClogP calculation, the Me-branch alpha to the CO group is not allocated a branch factor) cyclamen aldehyde heptanal, 6-methoxy-2,6-dimethyl-

ClogPcalc

CAS

chemicals

Table 2. Aliphatic Aldehydes Investigated and Their Calculated Physicochemical and Reactivity Parametersa

Chemical Research in Toxicology Article

DOI: 10.1021/acs.chemrestox.7b00050 Chem. Res. Toxicol. 2017, 30, 1309−1316

Article

Chemical Research in Toxicology

Figure 1. Schiff Base aliphatic aldehyde electrophiles.

where D is the dose of sensitizer; k is the rate constant or relative rate constant for the reaction of the sensitizer with a model nucleophile, and P is the 1-octanol/water partition coefficient quantifying hydophobicity. Equation 1 models competition between the carrier haptenation reaction in a hydrophobic environment and removal of the sensitizer through partitioning into polar lymphatic fluid. At present, experimental in chemico reactivity parameters for SB electrophiles are not available. As reported in the literature,33 a number of in chemico reactivity assays have been developed for skin sensitization, in particular the direct peptide reactivity assay (DPRA),34 which measures the depletion, in aqueous solution, of a peptide (with either a cysteine or a lysine nucleophilic center) by the chemical being tested. A kinetic variant of the cysteine based assay, enabling rate constants to be measured, has been developed.22 Although many SB electrophiles cause cysteine−peptide depletion in the DPRA,34 this is often due to oxidation rather than to binding and does not give a reliable measure of electrophilic reactivity.35 The lysine version of the DPRA is unsuitable for the SB reaction;34 this is because the DPRA has of necessity to be carried out in dilute aqueous solution, and the presence of large excess of water can suppress the SB reaction (which is reversible and produces water).36 Without adequate experimental in chemico reactivity parameters, for SB electrophiles, modellers have looked for appropriate substituent-based parameters. The Taft σ*

Figure 2. Schiff Base aromatic aldehyde electrophiles. molecular weight, which was used as the dependent variable. For the aromatic aldehydes, a QMM cannot be generated, but potency predictions are made by using the QMM for aliphatic aldehydes.



RESULTS AND DISCUSSION The RAI concept25 continues to be a useful approach in analyzing experimental sensitization potency data. In its general form, the RAI is expressed as RAI = log D + a log k + b log P

(2) 1312

DOI: 10.1021/acs.chemrestox.7b00050 Chem. Res. Toxicol. 2017, 30, 1309−1316

Article

Chemical Research in Toxicology

Figure 3. pNOEL observed vs pNOEL calculated from eq 3. 1,2,3,4,5,6,7,8-Octahydro-8,8-dimethyl-2-naphthaldehyde (triangle) is excluded from the regression. The chemical tested may have contained a potent hydroperoxide as an autoxidation-derived impurity.

pNOEL values can be compared against the values calculated from eq 3 for each of the 20 individual chemicals. With one exception (see below), the difference between observed and calculated pNOEL values is less than 0.35 log units, corresponding to a factor of