Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Danger signals in a rat model of nevirapine-induced skin rash
by
Xiaochu Zhang
A thesis submitted in conformity with the requirements
for the degree of Doctor of Philosophy
Graduate Department of Pharmaceutical Science
University of Toronto© Copyright by Xiaochu Zhang 2012
ii
Danger signals in a rat model of nevirapine-induced skin rash
Xiaochu Zhang
Doctor of Philosophy
Graduate Department of Pharmaceutical Sciences
University of Toronto
2012
Abstract
Nevirapine (NVP) can cause serious skin rashes and hepatotoxicity. It also causes an immune-
mediated skin rash in rats but not hepatotoxicity. There is strong evidence that the rash is due to
12-hydroxynevirapine (12-OH-NVP), which is further metabolized to a reactive benzylic sulfate
in the skin. This could both act as a hapten and induce a danger signal. In contrast, most of the
covalent binding in the liver appears to involve oxidation of the methyl group leading to a
reactive quinone methide. In this study we examined the effects of NVP and 12-OH-NVP on
gene expression in the liver and skin. Both NVP and 12-OH-NVP induced changes in the liver,
but the list of genes was different, presumably reflecting different bioactivation pathways. In
contrast, many more genes were up-regulated in the skin by 12-OH-NVP than by NVP, which is
consistent with the hypothesis that the 12-hydroxylation pathway is involved in causing the rash.
Some genes up-regulated by 12-OH-NVP were Trim63, S100a7a, and IL22ra2, etc. Up-
regulation of genes such as S100a7a, which is considered a danger signal, supports the danger
hypothesis. Up-regulation of genes such as the ubiquitin ligase and Trim63 are consistent with
protein-adduct formation. Up-regulation of IL-22ra2 gene suggests an immune response. These
results provide important clues to how NVP causes induction of an immune response, in some
cases leading to an idiosyncratic drug reaction.
iii
Acknowledgments
I attribute all my work to my husband Bo Shao and our kids, Cassey Shao and Bill Shao. I want
to say Thank You from my heart to my supervisor Dr. Uetrecht, from whom I have learned so
much in both science and life. I am very grateful to my committee members, Dr. Houry, Dr.
Pennefather, and Dr. O’Brien, for their patience and advice. I am also grateful to all my lab
mates and a lot of other people who have helped me in many different ways. The last but not the
least, I want to say Thank you to Connie (Mrs. Uetrecht) for her kindness and warmth.
iv
Table of Contents
Contents
Acknowledgments .......................................................................................................................... iii
Table of Contents ........................................................................................................................... iv
List of Abbreviations ..................................................................................................................... vii
List of Tables ................................................................................................................................ xiii
List of Figures .............................................................................................................................. xiv
1 Introduction-Nevirapine ............................................................................................................. 1
1.1 NVP-induced idiosyncratic reactions .................................................................................. 3
1.1.1 NVP-induced liver toxicity ..................................................................................... 4
1.1.1.1 Clinical characteristics and risk factors .................................................... 4
1.1.2 NVP-induced skin rashes ........................................................................................ 5
1.1.2.1 Clinical characteristics and risk factors .................................................... 5
2 Definition and Characteristics of Different Types of IDR ......................................................... 7
2.1 Types of IDRs ..................................................................................................................... 8
2.1.1 Idiosyncratic drug-induced skin rash ...................................................................... 8
2.1.1.1 Skin histology ........................................................................................... 8
2.1.1.2 Maculopapular skin rashes ...................................................................... 10
2.1.1.3 Urticaria .................................................................................................. 12
2.1.1.4 DRESS .................................................................................................... 13
2.1.1.5 Fixed drug eruption ................................................................................. 13
2.1.1.6 SJS and TEN ........................................................................................... 14
2.1.2 Idiosyncratic drug-induced liver toxicity .............................................................. 15
2.1.3 Idiosyncratic drug-induced hematological adverse reactions ............................... 17
2.1.3.1 Drug-induced hemolytic anemia ............................................................. 17
2.1.3.2 Drug-induced thrombocytopenia ............................................................ 18
2.1.3.3 Drug-induced agranulocytosis ................................................................ 19
2.1.3.4 Drug-induced aplastic anemia ................................................................. 20
2.1.4 Idiosyncratic drug-induced autoimmunity ............................................................ 21
v
2.2 Mechanisms of IDRs ......................................................................................................... 24
2.2.1 Involvement of the immune system in the mechanism of IDRs ........................... 25
2.2.1.1 Hapten hypothesis ................................................................................... 26
2.2.1.2 Danger hypothesis ................................................................................... 27
2.2.1.3 Pharmacological interaction (p-i) hypothesis ......................................... 29
2.2.1.4 Immune balance ...................................................................................... 31
2.2.1.5 Mitochondrial damage ............................................................................ 31
2.2.1.6 Viral reactivation .................................................................................... 32
2.2.1.7 Epigenetic effects .................................................................................... 33
2.2.1.8 Direct activation of antigen presenting cells ........................................... 33
2.2.2 Involvement of reactive metabolites ..................................................................... 34
3 Animal models .......................................................................................................................... 43
3.1 Penicillamine-induced autoimmunity in rats ..................................................................... 45
3.2 Sulfonamides in dogs ........................................................................................................ 48
3.3 Propylthiouracil-induced lupus in cats .............................................................................. 48
3.4 NVP-induced skin rash model in rats ................................................................................ 49
3.5 Danger signals in NVP-induced skin rash ......................................................................... 60
3.5.1 Danger signals in IDRs.......................................................................................... 60
4 Hypothesis ................................................................................................................................ 63
4.1 Strategy .............................................................................................................................. 63
5 Materials and Methods ............................................................................................................. 65
5.1 Materials ............................................................................................................................ 65
5.2 Methods ............................................................................................................................. 66
5.2.1 Animal Care .......................................................................................................... 66
5.2.2 Drug administration ............................................................................................... 66
5.2.3 Synthesis of 12-OH-NVP ...................................................................................... 67
5.2.4 Synthesis of NDVP ............................................................................................... 68
5.2.5 Mass spectrometry ................................................................................................. 68
5.2.6 Microarray study of rat liver, skin, whole ear and ear skin ................................... 69
5.2.7 Immunohistochemistry .......................................................................................... 70
5.2.8 Synthesis of rabbit anti-rat S100a7a antibody ....................................................... 71
5.2.9 Western blotting .................................................................................................... 71
5.2.10 2D-electrophoresis................................................................................................. 72
vi
5.2.11 ELISA analysis ...................................................................................................... 73
5.2.12 Real time-PCR ....................................................................................................... 73
6 Results ...................................................................................................................................... 76
6.1 Microarray analysis of gene expression changes in the whole ear tissue or peeled ear tissue after NVP, 12-OH NVP, or DNVP treatment for 6 or 12 h .................................... 76
6.2 Real-time PCR and protein level study of some genes in the ear and serum .................... 82
6.3 Changes in gene expression in the liver 6 or 12 h after NVP or 12-OH-NVP treatment .. 91
6.4 Changes in gene expression in the skin after NVP or 12-OH-NVP treatment .................. 96
6.5 Blood levels of IL-22ra2 and S100a7a protein in the skin .............................................. 102
7 Discussion ............................................................................................................................... 105
References ................................................................................................................................... 114
Appendices .................................................................................................................................. 129
vii
List of Abbreviations
12-hydroxynevirapine 12-OH-NVP
2-hydroxynevirapine 2-OH-NVP
3-hydroxynevirapine 3-OH-NVP
aminobenzotriazole ABT
adverse drug reaction ADR
alanine aminotransferase ALT
acetaminophen APAP
antigen presenting cell APC
antioxidant response element ARE
aspartate aminotransferase AST
basement membrane zone BMZ
bovine serum albumin BSA
cluster of differentiation CD
CCAAT/enhancer binding protein (C/EBP) delta Cebpδ
viii
3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate CHAPS
cytochrome P450 CYPs
cysteine Cys
3, 3’-diaminobenzidine DAB
delayed drug-induced hypersensitivity reaction DHRs
Drug-induced hypersensitivity syndrome (DIHS) DIHS
dimethyl sulfoxide DMSO
deuterated nevirapine DNVP
drug reaction with eosinophilia and systemic symptoms DRESS
dithiothreitol DTT
enhanced chemiluminescent ECL
endoplasmic reticulum ER
false discovery rate FDR
FK506 binding protein 5 Fkbp5
glyceraldehyde 3-phosphate dehydrogenase GAPDH
gene expression omnibus GEO
ix
reduced glutathione GSH
hour h
highly active antiretroviral therapy HAART
human constitutive androstane receptor hCAR
human herpes virus HHV
human leukocyte antigen HLA
high mobility group box 1 protein HMGB1
high-performance liquid chromatography HPLC
human pregnane X receptor hPXR
horseradish peroxidase HRP
heat shock proteins HSPs
intraperitoneal injection i.p.
half maximal inhibitory concentration IC50
intracellular adhesion molecule-1 ICAM-1
idiosyncratic drug-induced liver injury IDILI
idiosyncratic drug reaction IDR
x
isoelectric focusing IEF
interferon-gamma IFN-γ
immobilized pH gradient gel IPG
intravenous immunoglobulin IVIG
keyhole limpet hemocyanin KLH
methylcellulose MC
major histocompatibility complex MHC
minute min
multiple reaction monitoring mode MRM
metallothionein 1a Mt1a
the National Center for Biotechnology Information NCBI
nuclear factor kappa-light-chain-enhancer of activated B cells NF-κB
non-nucleoside reverse transcriptase inhibitor NNRTI
nuclear receptor subfamily 4, group A, member 3 Nr4a3
non-steroidal anti-inflammatory drugs NSAIDs
nevirapine NVP
xi
3’-phosphoadenosine 5’-phosphosulfate PAPS
phosphate buffered saline PBS
isoelectric point PI
pharmacological interaction p-i
a polymer of inosine and cytosine poly-IC
pyrroline-5-carboxylate reductase Pycr1
receptor of advanced glycosylation end products RAGE
sodium dodecyl sulfate-polyacrylamide gel electrophoresis SDS-PAGE
Stevens-Johnson syndrome SJS
sulfamethoxazole SMX
sulfotransferase SULT
T cell receptor TCR
toxic epidermal necrolysis TEN
tumor necrosis factor-α TNFα
2-amino-2-hydroxymethyl-propane-1,3-diol Tris
tyrosine Tyr
xii
ζ-associated protein of 70 kDa ZAP70
beta-2 microglobulin β2M
xiii
List of Tables
Tables:
Table 1. A comparison of characteristics of NVP-induced skin rash in humans and female Brown Norway rats (adapted from (1)).
Table 2. Genes with apparent high fold, but statistically nonsignificant, changes in whole rat ear 6 h (column A) or 12 h (column B) after NVP treatment.
Table 3. Genes with apparent high fold, but statistically nonsignificant, changes in whole ear 6 h (column A) or 12 h (column B) after 12-OH-NVP treatment, or 6 h after NVP treatment (column C).
Table 4. Genes with apparent high fold, but statistically nonsignificant, changes in peeled ear skin 6 h after NVP treatment.
Table 5. Genes with apparent high fold, but statistically nonsignificant, changes in peeled ear skin after 6 h NVP (A) or DNVP (B) treatment.
Table 6. A comparison of the microarray data from the ear 6 h after NVP (A, taken from Table 2), 12-OH-NVP (B, taken from Table 3), or DNVP treatment (C, taken from Table 5).
Table 7. Genes with a statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h (A) or 12 h (B) after NVP treatment.
Table 8. Genes with a statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h (A) or 12 h (B) after 12-OH-NVP treatment.
Table 9. Examples of genes with a significant change in gene expression in the skin 6 h after 12-OH-NVP treatment.
Table 10. Examples of genes with a significant change in gene expression in the skin 6 h after NVP treatment.
Page:
51
77
79
80
81
83
93
94
99
100
List of Figures
Figures:
Figure 1. Management of rash during Viramune therapy (from the product monograph on use of Viramune in the treatment of adults and children with HIV infection) (2).
Figure 2. Histology of mouse and human skin adapted from (3).
Figure 3. An illustration of the major mechanistic hypotheses for immune-mediated idiosyncratic drug reactions.
Figure 4. Reactive cations (electrophiles) formed by the loss of SO42- (adopted from (4)).
Figure 5. Major metabolic pathways of NVP (adopted from (5)).
Figure 6. A proposed scheme of bioactivation and possible reactive metabolites of NVP (adapted from (6)).
Figure 7. Putative bioactivation pathways of NVP (adapted from (5)).
Figure 8. Three major oxidative metabolites of NVP: 2-OH-NVP, 3-OH-NVP and 12-OH-NVP. Replacement of the methyl hydrogens with deuterium (DNVP) decreases the formation of 12-OH-NVP.
Figure 9. A putative bioactivation pathway of NVP in the liver.
Figure 10. A putative bioactivation pathway of NVP in the skin.
Figure 11. Real time-PCR study of the expression of Mt1a, Mt2a, Fkbp5, and S100a7a mRNA in the ear after NVP treatment.
Figure 12. Real time-PCR study of gene expression (relative concentration) of Nr4a3 in rat ears 6, 24, or 48 h after NVP treatment (A), 6 or 12 h after 12-OH-NVP treatment (B) or 6 h after NVP treatment (C).
Figure 13. The top panel summarizes the microarray data of S100a7a gene expression in the ear 6 or 12 h after NVP treatment. The bottom panel is western blotting analysis and immunohistochemistry analysis of S100a7a protein in the ear after NVP treatment.
Page:
2
10
34
42
52
53
54
56
57
58
84
85
87
xv
Figure 14. A summary of microarray analysis (top panel) of HMGB1 gene expression in the ear 6 h after NVP treatment and real time-PCR analysis (bottom panel) of HMGB1 gene expression in the ear 6, 12, 24, 48, or 72 h after NVP treatment or in control ears.
Figure 15. Western blotting and 2D-electrophoretic analysis of HMGB1 protein in the ear after NVP treatment.
Figure 16. ELISA analysis of the HMGB1 protein concentration (ng/mL mean ± s.d, n=4) in rat serum 6 or 12 h after NVP, 12-OH-NVP or DNVP treatment or in control rats.
Figure 17. Comparison of amino acid sequence between human CYP2B6 and Brown Norway rat CYP2B1 proteins.
Figure 18. A: Clustering of 525 genes from the one-way ANOVA analysis for statistically significant genes among three drug (NVP, 12-OH-NVP, and MC control) treatment groups (p value of treatment with FDR < 0.05) in 12 skin samples; B: A summary of a further one-way ANOVA analysis for the p value among the three drug treatment groups in 10 skin samples (one sample taken out from each NVP and 12-OH-NVP treatment group).
Figure 19. A. Fold changes in gene expression in the skin 6 h after 12-OH-NVP or NVP treatment. B. The pathway analysis of genes with changes in expression after 12-OH-NVP treatment using Ingenuity software.
Figure 20. The top panel is the serum level of IL-22ra2 in rats after NVP or 12-OH-NVP treatment. B. The bottom panel is the serum level of NVP, 12-OH-NVP metabolite (from NVP treatment) and 12-OH-NVP in rats after NVP (n=2) or 12-OH-NVP (n=4) treatment in food in the same experiment.
Figure 21. Western blotting analysis of S100a7a expression in rat skin after NVP (n=4) or 12-OH-NVP (n=4) treatment in food for 8 days.
88
89
90
95
98
101
103
104
1
1 Introduction-Nevirapine
Nevirapine (NVP, Viramune®), a drug for the treatment of HIV-1 infections, is a non-nucleoside
reverse transcriptase inhibitor (NNRTI). NVP was developed by Boehringer Ingelheim
Pharmaceuticals, Inc. and was the first NNRTI approved by the FDA in 1996. Although it is
commonly used in combination with other antiretroviral drugs in highly active antiretroviral
therapy (HAART, an antiretroviral regime of three or four drugs from different antiretroviral
classes) (7), patients require monitoring for adverse reactions, such as skin rash and liver
toxicity, the guidelines for which is shown in Figure 1.
NVP is a dipyridodiazepinone that selectively inhibits HIV-1 reverse transcriptase by directly
binding to the enzyme amino acid residues 181 and 188 (8-9). For wild-type reverse
transcriptase, the IC50 of NVP is 10.6 ng/mL (0.04 µM), while for the most common mutant
reverse transcriptase enzyme (Tyr-181 to Cys) the IC50 of NVP is 700 ng/mL (2.6 µM) (10). It
is commonly used in combination with other antiretroviral drugs in HIV-1 infection treatment to
overcome the selection of resistance, which is a major problem for anti-retroviral drugs (11). A
single dose is also useful in blocking HIV-1 transmission from mother to baby during labor and
the postnatal period (12-13). The pharmacokinetics and biotransformation of NVP were
characterized in early clinical studies (10, 14-16). NVP is a weak base (pKa = 2.8), and its
bioavailability is >90%. NVP is about 60% bound to plasma proteins in the plasma concentration
range of 1-10 µg/mL (archived drug label from Boehringer Ingelheim Pharmaceuticals, Inc.).
2
Figure 1. Management of rash during Viramune therapy (from the product monograph on use of
Viramune in the treatment of adults and children with HIV infection) (2).
The half-life of NVP after one dose (200-400 mg/day) is about 45 h, but it decreases to 25-30 h
after multiple dosing (17). The average peak plasma level of NVP is about 3.4 + 1.0 µg/mL 4 h
after the first dose (400 mg/day), while the average steady-state peak and trough concentrations
were 7.2 + 1.4 µg/mL and 4.0 + 1.2 µg/mL (higher than the IC50), respectively. The major route
of NVP clearance in humans is via liver metabolism, including P-450 oxidation and glucuronide
3
conjugation, while the major elimination route is urinary excretion of glucuronide metabolites
(11, 14). Only a small percentage of parent drug (2.7%) is excreted in urine (10, 14). The major
hydroxylated metabolites of NVP in humans are 3-, 12-, 2-, 8- hydroxy-NVP (OH-NVP) (14),
and the same metabolites also formed in rats (18).
An in vitro study also found that the major metabolites of NVP formed by human hepatic
microsomes were 2-, 3-, 8-, and 12-OH-NVP (19). The formation of 2- and 3-OH-NVP is
mediated by CYP3A4 and CYP2B6, while formation of 8- and 12-OH-NVP are mediated by
CYP2D6 and CYP3A4. An interesting finding in humans when dosed from 2.5 to 400 mg was
that the plasma concentration of NVP was not proportional to NVP dose, and the half-life of
NVP changed from 45 h after single dose to 30 h after multiple doses. This suggested that NVP
is an inducer of cytochromes P-450, and this was confirmed by an in vivo study in humans that
found that NVP induced both CYP3A4 and CYP2B6, which is consistent with a study in rats in
which NVP induced CYP2B1 and CYP3A (10). Although NVP is an inducer of CYPs, it was
also found to be an inhibitor of CYP3A4 (19); therefore, drug-drug interaction should be
considered in the clinical use of NVP. The most important interactions were found with
efavirenz, ketoconazole, rifampicin, and St John’s Wort. NVP decreased the level of efavirenz
and ketoconazole, while rifampicin and St John’s Wort decrease NVP levels (2).
1.1 NVP-induced idiosyncratic reactions
NVP was generally safe and well tolerated in early clinical trials with the major side effect being
skin rash (20). However, in 2000, the FDA issued a black box warning for NVP-induced life-
threatening hepatotoxicity and skin reactions.
4
1.1.1 NVP-induced liver toxicity
The incidence of NVP-induced hepatotoxicity in HIV patients was reported to be 1% in clinical
trials (20), but the incidence increased to 2.8% when more patients were exposed to NVP (21). In
the EuroSIDA database, NVP-induced liver failure was 0.3 case per 100 patient years, and 11 of
14 of these patients died of liver failure (22).
1.1.1.1 Clinical characteristics and risk factors
The warning from the NVP prescribing insert (published by Boehringer Ingelheim
Pharmaceuticals, Inc.) stated that severe, life-threatening, and in some cases fatal hepatotoxicity,
particularly in the first 18 weeks, has been reported in patients treated with NVP. NVP-induced
liver toxicity, which is characterized by elevated ALT, usually occurs within first 6 weeks, but it
may be delayed as late as 18 weeks (22). NVP-induced liver toxicity is also dose-dependent: 400
mg/day was associated with higher incidence of liver toxicity than 200 mg twice a day (23);
however, higher NVP plasma concentrations were not associated with a higher incidence of liver
toxicity (22). Liver injury with an elevated ALT is still an indication for discontinuation of NVP
treatment (2).
The risk factors for NVP-induced liver toxicity include female gender and higher CD4+ T cell
counts at initiation of therapy. Women with CD4+ T cell counts >250 cells/mm3, including
pregnant women receiving NVP in combination with other antiretrovirals for the treatment of
HIV-1 infection, are at greatest risk. However, hepatotoxicity associated with NVP use can occur
in both genders, any CD4+ T cell count, and at any time during treatment.
5
1.1.2 NVP-induced skin rashes
NVP-induced skin rash is the major adverse effect from NVP treatment (24).
1.1.2.1 Clinical characteristics and risk factors
NVP-induced skin rash is virtually always delayed in onset on first exposure, which is typical of
a drug-induced hypersensitivity reaction. Based on early clinical trials, 65% of NVP-induced
skin rash occurred in first 6 weeks (20). Severe skin rash, such as Stevens-Johnson
syndrome/toxic epidermal necrolysis syndrome (SJS/TEN), started from 10-240 days (median,
12 days) after the start of NVP treatment (25).
When NVP was dosed at 400 mg/day the incidence of skin rash was about 48%. The incidence
of NVP-induced skin rash was higher in patients with higher CD4 T cell counts (24). After the
treatment regime was changed to a two week lead-in low dose treatment of NVP 200 mg/day, the
incidence of skin rash was reduced to 18% (10). Another controlled trial showed that the
incidence of NVP-attributable rash was 16%, of which 65% developed a rash within the first 6
weeks of therapy. This study also showed that the lower lead-in dose (200 mg/day vs 400
mg/day) for the first 2 weeks reduces the frequency of drug-associated rash. However, NVP-
induced rash did not correlate well with NVP plasma levels (11). Serious rash, e.g. SJS/TEN,
occurred with an incidence of 0.3% (20). Severe rashes requiring drug discontinuation occur
with an overall incidence of 6%.
The incidence of NVP-induced skin rash was also gender-related. In a study of sex differences in
NVP-induced skin rash, woman were more susceptible to NVP-induced skin rash; they had a 7-
6
fold increase in risk for severe rash and were 3.5 times more likely to discontinue NVP therapy
(26). This finding was also confirmed in another study in Chinese patients (27).
Most NVP-induced skin rash manifests as a diffuse maculopapular rash or erythematous rash,
with or without constitutional symptoms (24), and can be classified as mild or severe (22). Mild
rash indicates a rash with intact skin and no systemic signs e.g. fever, lymphadenopathy, or
elevated hepatic transaminases; severe rash indicates severe erythema, skin blistering, erythema
multiforme, etc. plus the aforementioned systemic signs. Severe rash can manifest as SJS or TEN
with an incidence of about 0.3% (20).
NVP rechallenge in people who have a history of NVP-induced skin rash is dangerous. In one
study, people who developed NVP-induced skin rash were rechallenged with either NVP or
delavirdine (28). Most patients developed a rash after rechallenge with NVP, while 70%
developed rash after rechallenge with delavirdine. These recurrent rashes were more severe and
had a rapid onset. Therefore, there is a warning against rechallenge of NVP in the NVP-
prescribing insert (published by Boehringer Ingelheim Pharmaceuticals, Inc.): “Viramune
should not be restarted following severe skin rash; skin rash combined with increased liver
enzyme levels or other constitutional symptoms; or a hypersensitivity reaction. Liver function
tests should be performed if patients present with a suspected NVP-associated rash. Patients with
rash-associated ALT or aspartate aminotransferase (AST) elevations should permanently
discontinue Viramune therapy. Fatal NVP-induced hepatotoxicity and skin rash have also been
reported in prophylaxis cases” (29).
7
2 Definition and Characteristics of Different Types of IDR
NVP-induced liver toxicity and skin rash represent idiosyncratic drug reactions (IDRs).
Idiosyncratic means specific for an individual (30). An IDR is an adverse drug reaction that does
not occur in most patients who take a drug within the therapeutic range, and also does not
involve the pharmacological effects of the drug (30). IDRs are also referred to as type B (bizarre)
adverse drug reactions (30). Overall, ADRs are a major cause of patient morbidity and mortality
(31). Although IDRs only make up about 5% of ADRs, given the large variety of drugs that
cause IDRs and the number of people who take drugs, the number of cases is significant (32). In
addition, they can be very severe, e.g. idiosyncratic drug-induced liver injury was responsible for
nearly 13% acute liver failures in United States from 1997 to 2001 (33). Furthermore, the
unpredictability of IDRs makes it very unlikely that they will be discovered in clinical trials.
From 1975 to 2000, about 10% of new drugs approved in the US were either withdrawn or
received a black box warning due to unexpected IDRs (34). This uncertainty significantly
increases the overall cost of drug development. In addition, much of the preclinical testing is
performed to try to prevent IDRs, and although such testing is not very effective, it adds to the
time required for drug development, which further adds to cost.
IDRs are often characterized as being dose-independent. This is not true; in fact, the risk that a
drug will cause a significant risk of IDRs is related to the therapeutic dose of the drug. Drugs
given at a dose of less than 10 mg/day rarely cause IDRs (30), and 77% of 598 cases of
idiosyncratic drug-induced liver injuries were found to be from drugs given at a dose of greater
than 50 mg/day (35). What is true is that most patients will not have an IDR at any dose, and
there may not be any difference in incidence within the narrow range of usual doses of the drug.
8
In addition, the dose required to cause an IDR may be lower in a patient who has been previously
sensitized to drug, but there will always be a dose below which no one will have an IDR.
2.1 Types of IDRs
IDRs can affect any organ and take many different forms. Common types of IDRs include skin
rash, liver toxicity, hematological toxicity, and autoimmunity (36).
2.1.1 Idiosyncratic drug-induced skin rash
The most common type of IDR is skin rash. Many types of skin rashes can be induced by drug
administration, e.g. maculopapular rashes, urticaria, and SJS/TEN, etc. In order to understand
skin rashes it is important to understand the structure of the skin.
2.1.1.1 Skin histology
The skin is composed of three layers (37): epidermis, dermis, and hypodermis. The epidermis is
composed of stratified epithelium, which is arranged in continuous layers, i.e. (from bottom to
top) the basal layer (single layer), the Malpighian or prickle-cell layer or stratum spinosum (5-15
layers), the granular layer (1-3 layers) and the cornified layer (5-10 layers). The epidermis
renews itself continuously, and its major cell type is the keratinocyte (90-95%). Other cells in
epidermis include Langerhans cells, melanocytes, Merkel cells, and lymphocytes. The epidermal
appendages are hair follicles, sweat glands, and sebaceous follicles. The dermis is composed of
connective tissue with appendages such as vascular and nervous plexuses running through it. The
dermal-epidermal junction is a complex basement membrane synthesized by basal keratinocytes
and dermal fibroblasts. The hypodermis is composed mostly of subcutaneous fat called the
panniculus adiposus (37).
9
Because rats and mice are both rodents, a comparison of mouse and human skin histology
(Figure 2. adapted from a review (3)) will help to understand the basic differences between rat
and human skin. Both human and mouse skin have distinct compartments, e.g. epidermis,
appendages, dermis, etc. In both, the epidermis is composed mostly of keratinocytes and both
have various appendages, including hair follicles and sweat glands. Melanocytes are also present
to provide pigment to protect skin from UV damage and prevent photo-degradation of folate.
The dermis is under the epidermis and is composed of extracellular matrix, primarily fibroblasts,
vascular tissue, and immune cells (3). The innermost layer, which is beneath the dermis, is
subcutaneous adipose tissue. In contrast to these similarities, there are also significant differences
in the architecture of the skin compartments between mice and humans. Mouse epithelium has
much more densely-distributed hair follicles than that of human skin. Mouse hair follicles
undergo synchronous cycles during the first 2 months of life, while in human, follicles cycle
asynchronously (38). In addition, mouse epidermis is generally comprised of only 3 cell layers
and is <25 μm in thickness, while human epidermis is commonly composed of 6–10 cell layers
and is >100 μm thick (Figure 2) (3).
Therefore, drug absorption through mouse skin is greater than that in humans, which makes the
extrapolation of preclinical topical-drug delivery very difficult (39). Although both mouse and
human skin express NF-κB, the expression of other epidermal genes is different in mice and
human, e.g. activation protein 1 (40). Mouse skin also has a faster epidermal turnover and is
easier to transform using ultraviolet-light irradiation (41). Additionally, mice have a muscle layer
underlying all of the skin while humans do not (3).
10
Figure 2. Histology of mouse and human skin adapted from (3). The top panels are 5
magnification, while the bottom panels are 20 magnification. Samples were obtained from the
back of each species and demonstrate substantial differences, including epidermal and dermal
thickness, hair follicle density, dermal architecture, muscle layers, and location of melanocytes
(arrow, lower right panel). Specific tissue structures are labeled. Scale bars = 100 µm. B, basal
layer; BMZ, basement membrane zone; G, granular layer; S, squamous layer; SC, stratum
corneum.
2.1.1.2 Maculopapular skin rashes
Maculopapular, morbilliform, or exanthematous drug eruptions are probably the most common
type of skin rash and account for approximately 95% of all drug rashes. The appearance of
11
maculopapular rashes is “morbilliform” (resembling measles) with widespread fine
maculopapular pink or red to salmon-colored lesions that start on the upper trunk or the head and
neck. Maculopapular drug eruptions often spread symmetrically downward to the limbs in a
bilateral fashion and tend to become confluent. In addition, they can also manifest as a
scarlatiniform pattern: several pinpoint-sized pink to red papules may develop that coalesce and
give a sandpapery feel to the skin (42). Usually the rashes develop 1 to 2 weeks following the
initiation of the drug, and more rapidly on rechallenge or in previously sensitized patients
(43)(1). Maculopapular rashes are considered to be immune-mediated, specifically T cell-
mediated (44). Pathology studies reveal a cellular perivascular infiltration of T lymphocytes in
the dermis, consisting mostly of CD4+ T cells with fewer CD8+ T cells (44). Typically, an
interface dermatitis is present with varying degrees of accumulation of CD3+ and CD4+ T cells
(45). CD4+ T cells are mainly located in the perivascular dermis, whereas both CD4+ and CD8+
T cells are found at the dermo-epidermal junction zone. Both CD4+ and CD8+ T cells have been
shown to produce cytotoxic molecules such as perforin and granzyme B (46-47). Increased levels
of IFN-γ and TNF-α in the serum have also been reported.
Details of the initial steps in the initiation of drug-induced maculopapular rashes remain unclear.
Many of the mechanistic studies involve the lymphocyte transformation test where it was found
that the lymphocytes from many patients with a history of a rash proliferate in the presence of
the parent drug in a system in which there is no metabolism of the drug (45). This was
interpreted as demonstrating that an unreactive drug can bind reversibly to the antigen presenting
cell/T cell complex leading to the induction of an immune response. However, the unstated
assumption of this assay - what T cells respond to is what induced the immune response - was
proven wrong (48). Therefore the lymphocyte transformation test cannot be used to investigate
the initiation of the immune response. Despite this, the presence of drug-specific T cells
12
responding to the parent drug and/or its metabolites by proliferation and cytokine production
provide strong evidence for a T cell-mediated mechanism and can help to determine which drug
is responsible if the patient was on more than one drug.
2.1.1.3 Urticaria
Drug-induced urticaria, commonly called hives, represents approximately 5% of all cutaneous
drug reactions and is the second most common form of skin eruption after exanthematous
reactions (42). Urticarial lesions are characterized by raised, itchy, red blotches or wheals that
are pale in the center and red around the outside (42). They are widely scattered on the body, but
they can also be accompanied by deeper swelling of submucosal tissues. When dermal and
subcutaneous tissues are involved it is called angioedema. Urticarial lesions often fade within a
few hours without a trace, but angioedema takes longer to resolve.
The major mechanism for drug-induced urticaria involves the hapten hypothesis. The best-
studied example is penicillin-induced urticaria. β-lactams are chemically reactive and can
covalently bind to proteins, eliciting the production of IgE against the hapten-modified protein
(49)(2). Sufficient IgE production will result in significant allergic reactions such as urticaria and
anaphylaxis (50). This reaction is clearly immune-mediated because it is mediated by IgE
antibodies specific for drug-modified protein as demonstrated by skin tests. However, what
remains unclear is why different patients have different responses to the β-lactam-protein adduct.
Some recent studies reported a genetic association between β-lactam allergies and IL-13 and/or
IL-4Rα polymorphisms (51). This also supports an immune mechanism. However, not all drug-
induced urticaria is mediated by antibodies. For example, the inhibition of kinin degradation
caused by angiotensin-converting enzyme inhibitors, altered arachidonic acid metabolism by
aspirin, and non-steroidal anti-inflammatory drugs (NSAIDs), as well as a receptor-mediated
13
release of histamine by opiates all involve a non-immune mechanism, at least not an adaptive
immune mechanism (52). Urticaria can also be precipitated by physical factors such as exercise
and cold (53).
2.1.1.4 DRESS
Drug reaction with eosinophilia and systemic symptoms (DRESS) is a serious drug-induced
hypersensitivity. The systemic symptoms include severe rash and fever. Sometimes DRESS is
lethal, especially when it overlaps with TEN and other drug hypersensitivity syndromes (54).
Other symptoms of DRESS include lymphadenopathy, arthralgias, and involvement of organs
such as liver, kidney, lungs, thyroid gland, bone marrow, and less commonly the brain (55). The
rash usually starts with a maculopapular rash. It has been estimated to occur in about one in
10,000 exposures with drugs such as anticonvulsants and sulfonamides (56). The list of drugs
commonly associated with this syndrome is similar to that which causes TEN, that is,
carbamazepine, phenytoin, sulfonamide antibiotics, minocycline, allopurinol, gold salts, and
dapsone. DRESS occurs with a higher incidence in people with African ancestry and usually
begins 2–6 weeks after initiation of treatment with the offending drug. Eosinophilia is common
in DRESS, and the rash and hepatitis may persist for several weeks after drug withdrawal. The
mortality rate of DRESS is about 10% (56). The mechanism appears to involve reactivation of a
herpes virus as discussed later.
2.1.1.5 Fixed drug eruption
Fixed drug eruptions are always drug-induced and always occur at the same site, although with
repeated exposure, the number of sites can increase (53, 57). In sensitized individuals, the lesions
usually occur in less than 2 days after re-exposure, and the most common site is on mucous
14
membranes such as the lips and external genitalia. Histologically, this rash is associated with a
dermal perivascular infiltrate of lymphocytes, eosinophils, and sometimes neutrophils. On
resolution, there is usually hyperpigmentation at the site, and macrophages associated with these
lesions contain melanin (55). The amnestic response of an isolated group of skin cells is most
easily explained by an immune mechanism (53).
2.1.1.6 SJS and TEN
SJS and TEN are two forms of life-threatening skin rashes, both characterized by fever, blister
formation, and differing only in the degree of severity. In SJS, epidermal detachment is less than
10% of the body surface area; TEN is more severe with involvement of ≥30% of the body
surface area and is associated with a >30% mortality. Involvement of between 10% and 30% is
termed transitional SJS-TEN (58). Histologically, both of SJS and TEN are characterized by
extensive keratinocyte apoptosis, which results in the separation of the epidermis from the
dermis, and this is believed to be mediated by CD8 T cells. One proposed mechanism is that the
interaction between soluble Fas produced by peripheral blood mononuclear cells and the Fas
ligand expressed on diseased keratinocytes initiates the extensive apoptosis of keratinocytes (59).
Another possible mechanism is through the release of cytotoxic mediators, such as perforin and
granzyme B. It is suggested that the elevated levels of TNF-α and Fas ligand originate from
keratinocytes, which may increase the expression of MHC I expression on keratinocytes, making
them more sensitive to cytotoxic cells (60).
Carbamazepine- and allopurinol-induced SJS/TEN are clearly associated with HLA-B*1502 and
HLA-B*5801, respectively (61-62). These genetic predispositions are drug-specific and vary
with ethnicity. The former was only found in some Asian populations (Han Chinese and a Thai
population) but not in Europeans, and the latter was found in both Han Chinese and Europeans.
15
The association between HLA and drug-induced SJS/TEN also provides evidence for an
immune-mediated mechanism.
NVP-induced SJS/TEN occurs with an incidence of about 0.3% (20), the treatment for which
should be immediate discontinuation of NVP (22). Symptomatic treatment with antipyretics,
antihistamines, or steroids is sometimes used, but no efficacy has been shown (22). Intravenous
immunoglobulin (IVIG) seems to be the most effective treatment but there are no randomized
trials.
2.1.2 Idiosyncratic drug-induced liver toxicity
Drug-induced liver injury (IDILI) is one of the most common serious IDRs. It represents a
significant impediment to drug development because it is the most common reason leading to
drug withdrawn from the market (63). It is also responsible for about 13% of all acute liver
failure in the United States (64). Although the mechanisms of IDILI are not well understood,
most IDILI appears to be caused by reactive metabolites, and presumably the reason that the
liver is a common target for IDRs is because it is the major site of drug metabolism. Although
drug metabolism leading to reactive metabolites is proposed to be involved in the pathogenesis
of most IDILI (30), there are no examples where polymorphism of a metabolic pathway is
sufficient to explain the idiosyncratic nature of IDILI. The involvement of reactive metabolites in
IDILI will be described in more detail in the next section.
The general hypothesis of why IDRs are idiosyncratic is that they are immune-mediated. There is
general consensus that most other types of IDRs are immune-mediated, but there is less
agreement in the case of hepatic IDRs. The evidence for immune-mediated mechanism includes
the delay between starting a drug and the onset of IDILI. The most typical delay is 1-3 months,
16
but in some cases, especially autoimmune IDILI, the delay can be more than one year (65). In
some cases there is a rapid onset of symptoms when a patient is rechallenged with a drug. In a
few cases, IDILI is associated with fever, rash, and eosinophilia, which are classic symptoms of
an immune-mediated allergic reaction (66). In addition, antidrug antibodies or autoantibodies
have been detected in some cases of IDILI (67). The histology of hepatocellular IDILI can mimic
viral hepatitis with mild to moderate inflammation and infiltration of mostly lymphocytes and
sometimes eosinophils (68).
However, many cases of IDILI are not associated with typical characteristics of immune-
mediated reactions, and this is the basis for the disagreement about involvement of the immune
system. Another hypothetical mechanism for IDILI is metabolic idiosyncrasy; however, as
mentioned above, there are no examples in which polymorphisms in drug metabolism are
sufficient to explain the idiosyncratic nature of IDILI (69). Although IDILI caused by drugs such
as isoniazid and ketoconazole are not usually accompanied by fever, rash, and eosinophilia and
often do not occur rapidly on rechallenge (66), there are clear cases of both isoniazid- and
ketoconazole-induced IDILI with a very rapid onset on rechallenge (70-71). This immune
memory provides strong evidence of an immune-mediated reaction.
In some cases of IDILI, there is a long delay in onset of the symptoms on rechallenge, which
indicates a lack of immune memory. For example, there is no recurrence on rechallenge in many
cases of isoniazid-induced IDILI, or they occur very late as with some cases of troglitazone-
induced hepatotoxicity. However, lack of immune memory does not mean that an IDR is not
immune-mediated. For example, in the case of heparin-induced thrombocytopenia, which is
clearly immune-mediated, there is also no immune memory. The delay in onset is actually
longest for IDILI that is clearly immune-mediated, i.e. for drug-induced autoimmune hepatitis.
17
For example minocycline can cause two different types of IDILI: one that is typical for IDILI
and occurs after 1-3 months of treatment, and the other that is autoimmune and occurs after more
than a year of treatment (65). Thus IDILI, especially the delay in onset, can most easily be
explained by immune mechanisms.
2.1.3 Idiosyncratic drug-induced hematological adverse reactions
The most common immune-mediated hematologic IDRs are hemolytic anemia,
thrombocytopenia, agranulocytosis, and aplastic anemia affecting red blood cells, platelets,
neutrophils, and all blood cells, respectively (72).
2.1.3.1 Drug-induced hemolytic anemia
Drug-induced immune hemolytic anemia is characterized by increased red cell destruction
through antibody-mediated complement activation (73). Three different types of antibodies:
hapten-specific antibodies, drug-dependent antibodies, and drug-induced autoantibodies, have
been associated with this IDR (72). A common drug leading to the formation of hapten-specific
antibodies is penicillin. Extensive penicillin treatment (high dose for more than 10 days) can
induce antibodies that bind to red cells and cause their destruction (74). In contrast to hapten-
specific antibodies, drug-dependent antibodies appear to modify specific red cell membrane
glycoproteins. The causative drugs, e.g. quinine, quinidine, and cefotetan, must be present for
hemolysis to occur, but the antibodies do not bind to the drug (72, 75). The third type of
antibodies that can induce hemolytic anemia are drug-induced autoantibodies, which bind to red
cells even when the causative drugs, e.g., α-methydopa, l-dopa, or procainamide, are not present.
Unlike the other forms of hemolytic anemia, which usually occur after a week or two of
treatment, the onset of autoimmune hemolytic anemia typically occurs only after 4-6 months of
18
drug treatment (72). However, only a small fraction of patients with positive antibodies had
significant hemolytic anemia (76). Clearly these are immune-mediated reactions, and anemia
mediated by hapten-specific antibodies supports the hapten hypothesis. The mechanism by which
drugs induce the other types of antibodies is not clear.
2.1.3.2 Drug-induced thrombocytopenia
Another common IDR is thrombocytopenia. The normal platelet count is above 150,000/µL of
blood, and when it falls below 10,000 platelets/µL the patient is at very high risk of life-
threatening hemorrhage. Drugs are a common cause of thrombocytopenia (77). Compared with
drug-induced immune hemolytic anemia, more drugs are involved in induction of
thrombocytopenia (72). A typical manifestation of drug-induced thrombocytopenia is
spontaneous bruising, and the time to onset is usually after a week or more of treatment with the
offending drug (78). A clear example of immune-mediated thrombocytopenia is caused by
heparin, which is mediated by antibodies against the heparin-platelet factor 4 complex (79). Even
though it is clearly immune-mediated, it is not associated with immune memory, which is a
common feature of immune-mediated reactions. Specifically, if a patient with a history of
heparin-induced thrombocytopenia is rechallenged with heparin they usually do not develop
thrombocytopenia, or if they do, it does not occur more rapidly (80). As in hemolytic anemia,
three types of antibodies: hapten-specific antibodies, drug-dependent antibodies, and drug-
induced autoantibodies, have been associated with the pathogenesis of drug-induced
thrombocytopenia (72). Penicillin is also the major cause for hapten-specific antibody, while
levodopa, procainamide, penicillamine, and sulfamethoxazole were implicated in the drug-
induced platelet-specific autoantibodies (81). More recently, biological drugs such as rituximab
(anti-CD20) and infliximab (anti-TNFα) have been associated with autoimmune
19
thrombocytopenia (72). Drug-dependent antibodies induced by quinine and many antibiotics
bind to glycoprotein (IIb/IIIa complex and GPIb/IX complex) on the platelet membrane to
induce platelet damage.
2.1.3.3 Drug-induced agranulocytosis
Another IDR is agranulocytosis, which is defined as a neutrophil count of less than 500 cells/µL
of blood. This places patients at high risk of infections. Most cases of agranulocytosis are
induced by drugs including analgesics, antipsychotics, antithyroid medications, and
anticonvulsants (82). Cancer chemotherapy can cause agranulocytosis that is not idiosyncratic,
but when it is idiosyncratic, in some cases there is evidence that it is immune-mediated (53, 83).
In one study, rechallenge experiments were performed on two patients with aminopyrine-induced
agranulocytosis, and after a single dose of aminopyrine, a precipitous drop in leukocyte count
occurred within 2 h (84). In another experiment, transfusion of blood from a patient with
agranulocytosis into a normal person who had just ingested aminopyrine resulted in a rapid drop
in neutrophil count (85). These experiments indicate that aminopyrine-induced agranulocytosis is
mediated by drug-dependent antibodies. In patients with agranulocytosis induced by the
antithyroid drug, propylthiouracil, antibodies that reacted with granulocytes, monocytes, and
hematopoietic precursor cells were detected (86); while in another report, antineutrophil
cytoplasmic antibodies against myeloperoxidase were detected in propylthiouracil-induced
agranulocytosis patients (87). In quinine-induced neutropenia, quinine-dependent neutrophil
antibodies react with the neutrophil’s several surface glycoproteins (88). However, there is also
drug-induced agranulocytosis in which the typical characteristics of an immune-mediated
reaction are not present (53, 72). The highly documented clozapine-induced agranulocytosis is an
example in which agranulocytosis does not recur rapidly on rechallenge (89), and no drug-
20
dependent antibodies have been reported in clozapine-induced agranulocytosis patients.
However, the lack of typical characteristics of an immune-mediated reaction does not prove that
clozapine-induced agranulocytosis is not immune-mediated as demonstrated by the example of
heparin-induced thrombocytopenia discussed above. Clozapine-induced neutropenia is related to
its metabolism in neutrophils where it is oxidized by myeloperoxidase to reactive nitrenium ion
metabolite which binds to neutrophils (90). The covalent binding between clozapine and
neutrophil proteins was detected in our lab (91), and these modified proteins have the potential to
initiate an immune response as mentioned in the discussion of the hapten hypothesis. Specific
HLA genotypes, which are important markers for susceptibility for clozapine-induced
agranulocytosis in Ashkenazi Jewish patients, were DRB1*0402, DQB1*0302, and
DQA1*0301, and in non-Jewish patients, HLA-DR*02, DQB1*0502, and DQA1*0102 (92).
However, the number of patients in the study was small (52 patients in total) and the associations
are relatively weak (53). Since the time to onset of clozapine-induced agranulocytosis is usually
6-12 weeks, and rechallenge of clozapine dose not shorten the time to onset, autoimmune
mechanisms may be involved (53). It has been shown that clozapine increases the rate of
apoptosis in vitro and leads to an increase in neutrophil turnover in vivo in rabbits without
leading to neutropenia (93). It is possible that in a few patients the neutrophil damage implied by
these results leads to an immune-mediated agranulocytosis.
2.1.3.4 Drug-induced aplastic anemia
The diagnosis of aplastic anemia is based on examination of the bone marrow in which most of
the hematopoietic cells have been replaced by fat, and this leads to a deficiency of all of the
blood cells described in the previous paragraphs (94). Although drug-induced aplastic anemia is
less common than drug-induced agranulocytosis, this severe adverse drug reaction has limited
21
the use of several drugs e.g., chloramphenicol and felbamate (95). It appears that idiosyncratic
drug-induced aplastic anemia is immune-mediated; specifically, mediated by cytotoxic T
lymphocytes that cause bone marrow destruction (96). Idiopathic aplastic anemia is sometimes
associated with viral infections (53), and it appears to be an autoimmune reaction (97-98). The
observation that both idiopathic and drug-induced aplastic anemia usually respond to
immunosuppressive therapy further supports an immune-mediated mechanism (94). It is still not
clear how this adverse reaction is initiated. Because drug-induced aplastic anemia and drug-
induced agranulocytosis can be induced by many of the same drugs, most of which can be
oxidized to reactive metabolites by the myeloperoxidase system of neutrophils, macrophages,
and some of their precursors (99), such bioactivation may be the common factor in these
hematological IDRs.
2.1.4 Idiosyncratic drug-induced autoimmunity
Autoimmunity is, by definition, an immune-mediated disease in which the immune system
attacks its own tissue and cells to induce damage. Drugs, such as hydralazine, procainamide,
isoniazid, α-methyldopa, quinidine, minocycline, and chlorpromazine are able to trigger
autoimmunity (100) in which the manifestations include autoantibodies, and a systemic lupus
erythematosus-like syndrome (36). Drug-induced autoimmunity is a good example of an
immune-mediated IDR.
The symptoms of drug-induced autoimmunity can be classified as generalized autoimmune
reactions that resemble idiopathic lupus and organ-specific autoimmune reactions such as
autoimmune hemolytic anemia and autoimmune hepatitis discussed above (101). Depending on
the type of reaction, patients may develop autoantibodies to nuclear antigens, to erythrocytes, or
22
to other protein antigens similar to idiopathic autoimmune diseases. These autoantibodies do not
disappear immediately after withdrawal of the offending drug, but the clinical symptoms usually
resolve within weeks even though, by definition, the autoantigen is still present (36).
Many drugs can induce generalized autoimmune syndromes: drug-induced vasculitis and a
lupus-like syndrome, which is generally a milder version of the idiopathic disorder and is usually
associated with production of antihistone antibodies (102). However, the clinical and serological
phenotypes of the drug-induced autoimmune reactions overlap with the idiopathic forms so that,
other than exposure to drug and resolution when the drug is stopped, it is hard to differentiate
them (103). Common clinical manifestations of drug-induced lupus are myalgias, arthritis, fever,
and serositis involving the pleura and/or pericardium (104). About 10% of lupus is estimated to
be drug-induced, with 15,000 to 30,000 cases occurring in the United States annually (100).
Similarly, about 10% of cutaneous vasculitis is reported to be drug-induced, with purpuric and
maculopapular rashes being the most common symptoms (105). Many drugs are suspected of
causing a lupus-like syndrome; it is difficult to determine an accurate number, but to date at least
38 medications have been implicated (102-103). Biological drugs such as anti-TNF-α antibodies
and cytokines such as interferon-α can also cause a lupus-like syndrome (106). It usually takes
more than 1 year of treatment with the offending drugs before the syndrome becomes clinically
evident, although antinuclear antibodies are detectable much earlier (107). The presence of
antinuclear antibodies is virtually a sine qua non for the diagnosis (53).
There are several hypotheses for the mechanism of drug-induced lupus-like syndrome. One
likely mechanism is the inhibition of DNA methylation. Some drugs can result in T-cell DNA
hypomethylation, leading to the activation of T cells and a lupus-like disease (108), either
through decreased ERK pathway signaling (hydralazine) or through inhibition of DNA
23
methyltransferase (procainamide) (109). Another proposed mechanism for drug-induced lupus is
the oxidation of a drug by macrophages or other antigen-presenting cells leading to the formation
of a reactive metabolite that binds to antigen presenting cells leading to their activation (110). A
few drugs, such as penicillamine, hydralazine, and isoniazid, react irreversibly with aldehydes on
APCs leading to their activation, and they are also associated with a high incidence of drug-
induced lupus (111-112). A third theory is that TNF-α inhibitors may shift the T-helper profile:
by blocking the Th1 cytokine TNF-α it may shift the immune system to a Th2 profile with the
production of autoantibodies and the development of lupus-like features (113).
An interesting phenomenon is that a large fraction of the drugs that cause IDILI also cause
autoimmune IDRs (69). One possible explanation for these observations is that self protein is
modified by most reactive metabolites, and if the dominant immune response is to a liver protein
it can lead to liver toxicity, and if it is to a skin protein, it can lead to a skin rash. Since the T cell
receptor repertoire is different for each individual, the dominant response will be different in
each patient (69). Although drug-induced autoimmunity usually resolves rapidly when the drug
is discontinued, this is not always the case and obviously the antigen is still present in an
autoimmune reaction. The autoimmune property of IDILI can explain why there is sometimes a
longer delay in onset and possibly lack of immune memory in IDILI. It could also explain why
IDILI could begin a month after the drug had been discontinued and the drug is no longer
present, or why it sometimes progresses after the drug was stopped (69).
Although the mechanisms of IDRs are not known, most IDRs appear to be mediated by reactive
metabolites. The cumulative evidence suggests that IDRs are immune-mediated, and there are
several hypotheses for immune-mediated IDRs, e.g. Hapten-hypothesis, Danger hypothesis, etc.
In order to test these hypotheses, we need animal models to perform in vivo studies.
24
2.2 Mechanisms of IDRs
Characteristics of IDRs include incidence, time to onset, dose dependence, adaptation/tolerance,
cross-reactivity, and genetic associations. The incidence of an IDR to any given drug is usually
low (<0.1%) (114), and they represent about 5% of total drug-induced ADRs. A major
characteristic of IDRs is the delayed time to onset (32): about one week or more on first
exposure; however, the typical delay is different for different types of IDRs and for different
drugs. Common maculopapular rashes usually occur after one to three weeks of therapy, but
drug-induced hepatitis most commonly occurs after one to two months of therapy (30). This is
typical of an immune-mediated reaction because it requires at least a week on first exposure for
the few T lymphocytes that recognize a specific immunogen to proliferate to sufficient numbers
to result in a clinically evident immune response. When a patient who has had an IDR is
rechallenged to the same drug there is usually, but not always, a more rapid onset of the IDR.
Another characteristic of IDRs is adaptation/tolerance. Specifically, a drug that causes severe
IDRs in a small number of patients usually causes a much higher incidence of mild, reversible
IDRs. For example, if a drug causes idiosyncratic liver failure in a small number of patients, it
usually causes a much higher incidence of increased transaminases, which is usually transient
and returns to normal despite continued treatment (30). If the IDR is immune-mediated it is
likely that the adaptation represents immune tolerance.
Cross-reactivity is another characteristic of some IDRs, especially for the aromatic
anticonvulsant drugs, such as phenytoin, carbamazepine and phenobarbital. If a patient has an
IDR to one of these drugs, the patient will have similar IDR to the other two drugs with an
incidence of 40-60% (115-116).
25
Other associations include gender, age, and disease state. Women have a higher incidence of
IDRs to many drugs such as halothane-induced hepatitis and clozapine-induced agranulocytosis
(117-118); however, this is not the case with all IDRs. The risk of drug-induced liver toxicity
usually increases with age for most drugs (30). Some infectious diseases, e.g. mononucleosis and
HIV infection, appeared to increase the risk of some IDRs (30).
Some genes are associated with an increased risk of a specific IDR (30). In general, studies that
look for a strong association between polymorphisms in drug metabolism and the risk of an IDR
have been negative (30). The genetic factors that have been found to be a very strong risk factor
for a few IDRs involve the immune system, especially MHC I and MHC II. For example, it
appears that hypersensitivity reactions to abacavir are associated with HLA-B*5701 allele (119-
120). However, even most patients who carry the HLA-B*5701 allele will not have a
hypersensitivity reaction if they take abacavir; therefore, other factors must be involved. These
characteristics suggest that most IDRs are immune-mediated (53).
2.2.1 Involvement of the immune system in the mechanism of IDRs
As described above, there is strong evidence that most IDRs are immune-mediated. This is
certainly true for anaphylactic reactions and drug-induced autoimmunity. There is also strong
evidence for skin rashes and generalized hypersensitivity reactions. The major disagreement
involves idiosyncratic liver toxicity as discussed above. If most IDRs are caused by reactive
metabolites and most IDRs are immune-mediated, then the question becomes – How do reactive
metabolites induce an immune response? There are several hypotheses that address this issue.
The major hypotheses are the hapten hypothesis and the danger hypothesis. These hypotheses are
not mutually exclusive, and the mechanism may be different for different drugs. Another
hypothesis, the pharmacological interaction hypothesis, does not require a reactive metabolite.
26
An immune-mediated reaction can also be induced by direct activation of antigen-presenting
cells, by alteration in immune balance, and by epigenetic effects. These later mechanisms may,
but need not, involve a reactive metabolite. There are also hypotheses for the mechanism of
IDRs that do not require the adaptive immune system: mitochondrial damage and the
inflammagen hypothesis. These hypotheses for non-immune mechanisms will be discussed in the
section on idiosyncratic liver toxicity, which is the organ toxicity to which they have been
applied.
2.2.1.1 Hapten hypothesis
In 1935, Karl Lansteiner found that small molecules, e.g. 2,4- dinitrochlorobenzene and p-
nitrosodimethylaniline, did not induce an immune reaction unless they were bound to large
molecules such as protein (121); this was the basis for the hapten hypothesis (122). The hapten
hypothesis states that small molecules covalently bind to proteins, and the modified proteins act
as antigen to induce a hypersensitivity reaction. Small molecules that bind to proteins leading to
an immunogenic protein are referred to as haptens (3,123). In order for a chemical or a drug to be
able to covalently bind to proteins, it needs to be chemically reactive. A good example of
covalent binding leading to a hypersensitivity reaction is penicillin-induced allergic reactions. A
characteristic feature of penicillin is a β-lactam ring, which is chemically reactive and does not
require metabolic activation in order to irreversibly bind to amino and sulfhydryl groups on
proteins. In some patients this leads to IgE antibody formation and an allergic IDR to the
penicillin-protein adduct (124). The hapten hypothesis is clearly true for penicillin-induced
allergic reactions because it is the anti-penicillin IgE antibodies that mediate the IDR.
27
However, most drugs that can induce hypersensitivity reactions, e.g. sulfonamides, are not
chemically reactive. Karl Lansteiner proposed in his 1935 paper that some molecules, e.g.
nitrosodimethylaniline, may change in the body to acquire the ability to covalently bind to
proteins (121), which heralded the role of biotransformation of chemicals for their covalent
binding to proteins. Later, the ADRs caused by several drugs were found to be related to the
formation of reactive metabolites that bind to proteins (125-126). A good example is halothane-
induced hepatotoxicity. Halothane is oxidized by cytochrome P450 to the reactive trifluoroacetyl
chloride, and antibodies were found against trifluoroacetyl chloride-modified protein in most
halothane allergic patients (127). However, unlike anti-penicillin antibodies, it is not clear that
antibodies against trifluoroacetylated proteins mediate halothane-induced liver injury. They do
indicate that halothane has induced an immune response, and even if these antibodies are not
pathogenic, it is likely that halothane-induced hepatotoxicity is immune-mediated.
2.2.1.2 Danger hypothesis
A classic principle of immunology is that the immune system responds to foreign material, and
this is consistent with the hapten hypothesis where the binding of hapten makes the protein
foreign and leads to an immune response. However, it was found that not all autoreactive T cells
are deleted, and there is no difference between how self and foreign proteins are presented to T
cells. Therefore, there must be some additional mechanism to control immune responses to
prevent widespread autoimmunity. It was discovered that activation of antigen presenting cells
leading to expression of costimulatory molecules such as B7 is required for activation of T cells.
The interaction between MHC antigen complex on antigen-presenting cells (APCs) and T cell
receptor (TCR) on T cell is referred to as signal 1, while the interaction between other molecules
such as B7 on APC and CD28 on T cells is referred to as signal 2 or costimulation. The immune
28
response is initiated when both signal 1 and signal 2 are present, and tolerance will be induced if
only signal 1 is present (128)(4).
Janeway proposed that signal 2 was mediated by toll-like receptors that recognize evolutionarily
conserved molecules on pathogens (129). In contrast, Matzinger proposed that it is molecules
released by damaged cells that activate antigen presenting cells and control immune responses,
and it has been found that many such molecules also bind to toll-like receptors. This is known as
the danger hypothesis (130). An alternative hypothesis for the role of reactive metabolites in the
pathogenesis of IDRs is that reactive metabolites or their covalent binding to proteins can
interrupt cellular functions and induce stress; this may lead to the release of danger signals from
stressed cells to trigger an immune response (128, 131). However, the hapten and danger
hypotheses are not mutually exclusive, and the danger hypothesis could explain why not all
drugs that covalently bind to protein are associated with a significant incidence of IDRs.
Specifically, unless the reactive metabolite not only modifies protein but also causes cell damage
it will not induce an immune response. This raises an important question: does the danger signal
have to be caused by the drug or can other forms of cell damage such as viral infections provide
the costimulation required to lead to an immune response, which in the presence of covalently
bound drug, can lead to an IDR? There are examples where a viral infection has been found to
increase the risk of an IDR (132). An obvious increase in ampicillin-induced IDRs was observed
in patients with mononucleosis (133), and HIV infected patients have an increased risk of
developing an IDR to sulfonamides and other drugs (134). Another example of a danger signal is
injury, and surgery appears to increase the risk of procainamide-induced agranulocytosis 10 fold
(135). However, most IDRs do not occur in patients with viral infections or other obvious
sources of a danger signal.
29
Some documented danger signals include high mobility group protein 1 (HMGB1), IL-1a,
cytosolic calcium binding proteins of the S100 family, heat shock proteins (HSPs), uric acid, etc
(136). HMGB1 is a non-histone nuclear protein, of which the identified receptors are receptors
for advanced glycation end products (RAGE) and toll-like receptors 2, 4, and 9 (137-138). An
interesting characteristic of HMGB1 is that it goes through posttranslational modification in
activated monocytes, which leads to its translocation from the nucleus to cytosol and further to
the extracellular matrix (139-140). LPS treatment results in hyperacetylated lysine, while TNFα
induces phosphorylated forms of HMGB1 (140). In drug-induced adverse reactions, high serum
levels of HMGB1 have been found in acetaminophen-induced liver toxicity in which HMGB1
may act as a proinflammatory factor to initiate an immune response (141).
S100 proteins also appear to act as danger signals, which interact with toll-like receptor 4 and
appear to play an important role in the development of autoimmunity (142). Some members in
S100 family, i.e. S100A8/A9 (143, 144 ), S100B (145), S100A4(146) were reported to be
secreted into extracellular space to exhibit cytokine-like function (147), supporting that S100
proteins are ‘good’ danger signals because they can be released to interact with cell surface
receptors on APCs. S100A7/A15 (psoriasin) in the epidermis was found to play an important
role in the pathogenesis of psoriasis (148). Another group of proteins that has been referred to as
danger signals is HSPs. However, not every member of this group of proteins can act as a danger
signal. While Hsp70 was found to be an endogenous danger signal in activating the effector
function of NK cells (149), Hsp27 was found to act as an anti-inflammatory protein (150).
2.2.1.3 Pharmacological interaction (p-i) hypothesis
Another hypothesis is the pharmacological interaction (p-i) hypothesis as proposed by Pichler
30
(151). In this hypothesis the parent drug acts as a superantigen to bind reversibly to the complex
formed by the complex of MHC II on antigen presenting cells and the T cell receptor on T cells
to initiate an immune response. This hypothesis was based on the observation that T cell clones
from patients with a history of IDRs to sulfamethoxazole were activated as measured by
proliferation when incubated with sulfamethoxazole in absence of drug metabolism (151).
Although the same observation was made with other drugs, sulfamethoxazole is a primary
aromatic amine, and virtually all primary aromatic amine drugs given at a dose of 100 mg/day or
more are associated with a significant incidence of IDRs (32). Presumably this is because
aromatic amines are readily metabolized to reactive metabolites. A key assumption of the p-i
hypothesis is that what lymphocytes respond to is what initiated the immune response. In an
immune-mediated skin rash induced by NVP in rats, we found that lymphocytes from these
animals respond to NVP better than the 12-hydroxy metabolite (12-OH-NVP) even though we
had shown that oxidation to 12-OH-NVP is required to induce a rash. Furthermore, T cells from
animals in which the rash was induced by treatment with 12-OH-NVP and the animals had never
been exposed to NVP still responded better to NVP (5). Therefore, the response of T cells is not
an accurate indication of what induced an immune response. In a more recent study of human T
cells from 3 patients with hypersensitivity to sulfamethoxazole, lymphocytes proliferation was
detected for sulfamethoxazole and both hydroxylamine and nitroso metabolites; however, more
antigen-specific T-cell clones were generated with the two metabolites than with the parent drug
sulfamethoxazole (152). An IDR that may involve the p-i mechanism is ximelagatran-induced
hepatotoxicity. Ximelagatran is structurally similar to a small peptide and does not appear to
form reactive metabolites. It may be able to initiate an immune response through a p-i-type of
interaction, and there is evidence that it binds reversibly to a specific MHC (153).
31
2.2.1.4 Immune balance
The immune system is highly regulated and balance is maintained by various arms of the
immune system. Many new biological drugs, e.g. antibodies and cytokines, have been developed
to treat autoimmune diseases such as multiple sclerosis (53). However, paradoxically, even
though most are used for the immunosuppressant effects, they can also induce autoimmunity. For
example, anti-tumor necrosis factor-α (TNFα) antibodies can induce various autoimmune
syndromes such as lupus and vasculitis (106), while an interleukin-1 receptor antagonist
(anakinra) and an anti-CD20 antibody (rituximab) induced psoriasis (154-155). Although the
mechanisms of these reactions are not clear, they are likely to be related to an altered balance of
the immune system (53).
2.2.1.5 Mitochondrial damage
Mitochondrial damage is another hypothesis for the mechanism of IDILI (36). Valproic acid and
perhexiline-IDILI are characterized by microvesicular steatosis and/or lactic acidosis (66),
indicating that the drug has compromised lipid and energy metabolism in mitochondria. It was
found that mice that are heterozygous deficient in mitochondrial superoxide dismutase developed
delayed-onset liver damage when treated with troglitazone, a drug that was removed from the
market because of IDILI (156). However, the injury was relatively mild, and other researchers
were not able to reproduce the injury (157).
Drugs can cause damage to mitochondrial DNA that leads to delayed and cumulative liver
damage in humans, but this toxicity is not idiosyncratic (158-159). The delay and cumulative
liver toxicity observed with mitochondrial DNA damage is due to the fact that mitochondrial
DNA does not have the same repair mechanisms as nuclear DNA (36). This toxicity is also
32
characterized by microvesicular steatosis and/or lactic acidosis, which is not a common feature
of IDILI and unlikely to result in liver failure (69). If a drug caused damage to mitochondrial
proteins instead of DNA, it should not be delayed and cumulative because of the relatively rapid
turnover of proteins. Milder mitochondrial damage may not directly lead to liver failure, but it
could act as a danger signal, and in some patients, this might lead to an immune response that
results in liver failure. Therefore, mitochondrial damage could be an important component to the
mechanism of IDILI even if it is not sufficient by itself to explain most cases of IDILI.
2.2.1.6 Viral reactivation
Drug-induced hypersensitivity syndrome (DIHS), also referred to as DRESS, is a severe
multiorgan hypersensitivity reaction. DIHS usually appears after 3–6-weeks of exposure to the
offending drug, e.g. anticonvulsants (160-162), allopurinol (163), dapsone, and minocycline
(161, 164). The main symptoms of DIHS that were summarized earlier in the DRESS section
include exanthematous eruption, sometimes with small pustules, facial edema, high fever,
systemic lymphadenopathy, leukocytosis, eosinophilia, etc. (165). Other systemic manifestations
such as pneumonitis, myocarditis, etc. are similar to some viral infections such as infectious
mononucleosis, suggesting the involvement of viral infections in the development of DRESS
(165). DRESS has been associated with reactivation of human herpes virus, e.g. HHV-6 (54,
166-167), HHV-7 (165), and cytomegalovirus (168) or Epstein–Barr virus (169).
Although the exact relationship between reactivation of virus and development of DRESS is not
clear, the virus may mediate most of the symptoms of DRESS. A study showed that drugs
stimulate the replication of herpes virus, e.g. Epstein–Barr virus, in Epstein–Barr virus-
transformed B lymphocytes (54).
33
2.2.1.7 Epigenetic effects
Epigenetic effects consist of changes in gene expression caused by modifications of chromatin
(e.g, histone acetylation) and DNA (e.g. cytosine 5-methylation) that do not involve changes in
the DNA sequence (170). The pathogeneses of some diseases such as cancer, autoimmune
diseases, and asthma have been associated with epigenetic changes. For example,
hypomethylation of the DNA in T cells is proposed to drive the autoimmune response in lupus
(171). Epigenetic effects, including methylation of DNA and histone deacetylation, may also be
potential mechanisms for IDRs (83). 5-Azacytidine, which is an anticancer drug and also used in
the treatment of myelodysplastic syndrome, may cause neutropenia via DNA hypomethylation-
induced apoptosis (172).
2.2.1.8 Direct activation of antigen presenting cells
Many drugs are oxidized to reactive metabolites by the myeloperoxidase system in neutrophils
and macrophages including antigen presenting cells (99). Such reactive metabolites can lead to
the activation of antigen presenting cells (173). Most of these drugs cause autoimmunity and
other types of IDRs (99). In addition, it was found that one of the interactions between antigen
presenting cells and T cells involves a reversible imine bond produced by an aldehyde group on
the antigen presenting cell and an amino group on the T cell (174). We have found that
penicillamine and drugs that contain a hydrazine group such as hydralazine and isoniazid can
bind irreversibly to aldehyde groups on macrophages and lead to their activation (111-112).
These drugs are also associated with a high incidence of drug-induced autoimmune reactions.
The major mechanistic hypotheses for immune-mediated idiosyncratic drug reactions are
summarized in Figure 3.
34
Figure 3. An illustration of the major mechanistic hypotheses for immune-mediated idiosyncratic
drug reactions. APCs, antigen presenting cells; MHC, major histocompatibility complex; TCR,
the T cell receptor.
2.2.2 Involvement of reactive metabolites
A fundamental question in the pathogenesis of IDRs is whether they are caused by the parent
drug or their reactive metabolites; much circumstantial evidence strongly supports the idea that
most idiosyncratic drug reactions are due to reactive metabolites of drugs (175). Idiosyncratic
reactions to chemicals or drugs, i.e. hypersensitivity reactions, have been studied for long time.
In 1938, Fieser postulated that the polycyclic hydrocarbons might be converted in vivo to
derivatives that are able to form conjugates with tissue constituents (176). From the 1930s to the
1950s, some compounds, e.g p-dimethylaminoazobenzene, were used to test the hypothesis that
small organic molecules undergo bioactivation in animals to form electrophiles, which were able
35
to bind to macromolecules in tissue to induce toxicity (177). A remarkable finding was made
with the hepatic carcinogen, p-dimethylaminoazobenzene, by Miller and Miller, who found that
aminoazo dyes (metabolites of p-dimethylaminoazobenzene) tightly (covalently) bound to
proteins in the liver when rats were fed p-dimethylaminoazobenzene. This was the first in vivo
evidence that showed covalent binding between chemical metabolites and liver proteins and also
the likely connection between the covalent binding and carcinogenicity in the liver (178). All
these findings provided the foundation for the connection between the covalent binding that
occurs between chemically reactive derivatives and tissue proteins and toxicity (177). From the
1970s - 1980s, a group of researchers at the National Institutes of Health - Brodie, Mitchell,
Gillette, and Boyd - examined the correlation between bioactivation of a wide variety of small
organic molecules and organ toxicity. The examined molecules included 4-ipomeanol,
acetaminophen (APAP), halothane, isoniazid, furosemide. More recently, bioactivation and
covalent binding of clozapine (91), and bromobenzene (179) were also investigated. The
conclusion was that there may be a correlation between reactive metabolites, covalent binding,
and tissue damage (177). Many drugs have now been studied for bioactivation and toxicity, and a
table of examples (aminopyrine, amodiaquine, clozapine, imipramine, etc.) that undergo
bioactivation, form reactive metabolites, and induce toxicity can be found in a paper from Japan
(180).
Reactive metabolites are generally electrophiles or free radicals (181). An electrophile is a
molecule that is electron deficient and reacts with nucleophiles, which usually have a negative
charge or a lone pair of electrons that can form a bond to the electrophile. Factors that can result
in a reactive metabolite are a good leaving group, ring strain, a double bond conjugated with a
carbonyl group (Michael acceptors), and the presence of electron-withdrawing groups.
36
Good leaving groups are usually strong acids when protonated (181). The reason for a strong
acid to be a good leaving group is its ability to accept a negative charge on loss of the acidic
proton. For example, chloride and sulfate are good leaving groups and hydrochloric acid and
sulfuric acid are strong acids. Another example is aminobenzotriazole (ABT) whose oxidation
produces two molecules of nitrogen gas (a good leaving group) and benzyne. Benzyne is an
alkyne in a ring structure (strained cycloheptyne), which makes benzyne very reactive and it
readily covalently binds to P450. This covalent binding inactivates the P450 that oxidized ABT,
which is the mechanism that makes ABT one of most effective general P450 inhibitors (182).
Ring strain also increases the reactivity of a compound (181). The normal bond angle of a sp3-
hydridized carbon is 109o; therefore, a carbon in a three-membered ring in which the bond angle
is forced to be 60o is under a considerable amount of strain and a reaction that opens the ring is
facilitated (181). For example, an intramolecular reaction of mechlorethamine leads to an
aziridinium ion that is both positively charged and has ring strain. When it reacts with a
nucleophile, the ring strain will be relieved.
Michael acceptors are the alkenes that are polarized by conjugation with a carbonyl group, which
makes them reactive (181). A simple example is acrolein. A metabolite of felbamate is
phenylacrolein, which is presumably responsible for the adverse reactions of this drug (181).
Another type of activated double bond is found in isocyanates and isothiocyanates, e.g.
methylisocyanate, which can react with nucleophiles (181). Oxidation of tolbutamide (a
sulfonylurea) and troglitazone (a thiazolidinedione) to form isocyanates, which are reactive and
may be responsible for IDRs. Troglitazone was withdrawn due to its liver toxicity.
37
Carbenes are a unique type of reactive metabolite, a divalent carbon, which is a putative
intermediate in the oxidation of methylenedioxy-containing compounds (181). The resulting
carbene binds to the heme iron of P450 leading to its inactivation, which is also the mechanism
by which compounds can act to synergize insecticides by inhibiting their inactivation by insect
oxidative enzymes.
Free radicals are compounds with an unpaired electron (181). They are highly reactive,
especially those having elements such as O, N, C, etc. Since normal chemical bonds consist of
two electrons, free radicals do not generally covalently bind to nucleophiles. The major reaction
of free radicals is with other radicals, or they can abstract a hydrogen atom from a neutral
molecule (vitamin E and C, or unsaturated lipids) to generate a new, generally less reactive
radical (vitamin E and C free radicals or lipid free radicals) or abstract an electron to form an
anion to generate a radical cation. One electron oxidation can also form radicals. For example,
one electron oxidation of a cyclopropyl amine leads to ring opening and the formation of a
carbon-centered free radical and an iminium ion.
Since reactive metabolites react with nucleophiles (protein or DNA) to induce toxicity, it is
important to predict which drug or new chemical entity is likely to form reactive metabolites
(181). However, it is difficult to predict all potential reactive metabolites, and almost all drugs
have the potential to form a reactive metabolite. Some structures such as aryl amine/aryl nitro
groups, thiophenes, furans, and 3-methylindoles are associated with reactive metabolite
formation and are called ‘structural alerts’. Although not all drugs containing ‘structural alerts’
are associated with significant toxicity, these structures should generally be avoided.
One of the most common methods for detecting bioactivation of chemicals is by using small
molecule trapping agents, e.g. reduced glutathione (GSH) or cyanide, to form adducts with
38
reactive intermediates, which can be identified by LC-MS/MS or NMR. Another method is by
using radiolabeled compounds to form covalent binding with the liver proteins (in vivo) or
microsomes (in vitro). The radiolabel makes it possible to quantify the amount of covalent
binding by liquid scintillation counting and protein analysis for determination of radioactivity
and protein content, respectively. When radiolabeled compounds are fed to animals, the liver can
be analyzed to quantify in vivo covalent binding, while incubation of radiolabeled compounds
with microsomes can be used to quantify in vitro covalent binding.
Microsomes are composed of the endoplasmic reticulum (ER) from eukaryotic cells. They are
usually made from homogenized liver tissue, and metabolic enzymes, in particular cytochrome
P-450s (CYPs), are enriched in them. When these microsomes are incubated with chemicals,
metabolism and covalent binding between metabolic intermediates and CYPs can be
investigated. As mentioned before, the liver is presumably susceptible to IDILI because the liver
is the major metabolic organ. Covalent binding between reactive metabolites and liver proteins
may induce liver damage and interrupt liver function.
Twenty one drugs that either have been withdrawn from the US market or received a black box
warning from the FDA were reviewed for drug-induced hepatotoxicity in humans (114), and 5 of
6 withdrawn drugs and 8 of 15 drugs with a black box warning were found to form reactive
metabolites. A high daily dose, which increases the amount of reactive metabolite that can be
formed, has also been found to be a significant risk factor for the potential of a drug to cause
drug-induced idiosyncratic livery injury. Another study that compares drugs from four
categories (safe, warning, black box warning, and withdrawn) using radiolabeled drugs to
quantify covalent binding in three test systems (rat liver microsomes, human liver microsomes,
hepatocyte culture, and in vivo studies in rats), showed that daily dose and covalent binding were
39
correlated with the risk of idiosyncratic liver toxicity. In particular, covalent binding in
hepatocytes showed a significant correlation with the risk of idiosyncratic liver toxicity (183).
In a study of covalent binding and tissue distribution/retention of drugs that are associated with
idiosyncratic drug toxicity, it was found that higher covalent binding in human liver microsomes
was associated with more of the "problematic" drugs, including "withdrawn" and "warning"
drugs, than the "safe" drugs (180). In addition, the tissue distribution/retention of the drugs was
also examined by in vivo autoradiography to detect the residual radioactivity in the rat liver
observed at 72 or 168 h post-dose, which can be used for assessment of in vivo covalent binding
to liver proteins. Long-term (72-168 h) retention of radioactivity in the bone marrow was
observed with some drugs associated with agranulocytosis, e.g. amodiaquine and clozapine,
suggesting an association between the toxicity profile and drug distribution/retention (180). It is
interesting to see the consistency of covalent binding and tissue distribution/retention of various
“problematic” drugs with their ability to form reactive metabolites. These are studies of drugs
that have already been put on the market and cause toxicities. For development of new chemical
entities it was proposed that covalent binding assessment should be done as early as possible so
that the covalent binding potential can be designed out of the structure (177).
The question could be asked: “How much apparent covalent binding is acceptable in deciding
whether to advance a drug candidate into development?”. Merck Co. Inc, developed a “decision
tree” for assessing the suitability of lead compounds based on metabolic activation (177). In
brief, radiolabeled drug candidates are either incubated with human liver microsomes or rat liver
microsomes at a concentration of 10 µM for 1 h or fed to rats at a dose of 20 mg/kg, orally, and
the liver and plasma are taken after 2, 6, and 24 h. Then, covalent binding is quantified. If
covalent binding is less than 50 pmol eq./mg of protein both in vitro and in vivo, the candidates
40
can advance. On the other hand, if covalent binding is more than 50 pmol eq./mg protein in vitro
and/or in vivo, the candidate will be assessed based on qualifying considerations, e.g. the
potential to modify the structure and the availability of existing treatments, etc.
Because the liver is the major metabolic organ for drug bioactivation and reactive metabolites
may be too reactive to escape the liver, covalently binding to the liver enzymes is usually the
dominant site of binding (184). It is not hard to understand the correlation between covalent
binding and liver toxicity. Alternatively, some drug metabolites are stable enough to circulate to
the skin and covalently bind to skin proteins to induce hypersensitivity (185). Another possibility
is that drug metabolism also occurs in the skin, although skin is not a major metabolic organ.
Some drug metabolism does occur in the skin and the reactive intermediates bind to skin protein
to induce hypersensitivity (186).
Several types of cells in the skin (keratinocytes, fibroblasts, Langerhans cells, and melanocytes)
express both phase I and II enzymes, as well as transporters (187-188). In 1998, Cross et al. used
a microdialysis technique in human subjects, and they were able to demonstrate the metabolism
of methyl salicylate to salicylate in vivo after topical application (189). Other drugs, such as
dapsone, SMX, and phenytoin, have also been shown to be bioactivated in skin cells in vitro
(190-192). N-acetyl metabolites were detected for p-aminobenzoic acid, dapsone, and
sulfamethoxazole in cultured keratinocytes and dermal fibroblasts, which is consistent with the
expression of NAT1 mRNA in these cells (191, 193). Bioactivation of carbamazepine in humans
was detected by skin biopsy (194).
The mRNAs of many phase I enzymes are expressed in human skin as summarized in a review
paper (195); however, only flavin monooxygenase, CYP 1A1, 2B6, 2E1, and 3A4 proteins were
41
detected. Some phase II enzyme proteins, especially sulfotransferase 2B1 (SULT 2B1), has also
been detected in human skin (195).
One notable example of skin metabolism is the sulfation of minoxidil (196). When this drug was
used to treat hypertension, it promoted hair growth in some patients. It was found that sulfation
of minoxidil is important for hair follicle stimulation (197). Studies in skin biopsies and cultured
keratinocytes demonstrated that skin cells are capable of metabolizing minoxidil to its active
sulfate metabolite (196, 198-199).
SULTs are able to perform sulfonation (also referred to as sulfation) of hydroxyl and amine
substrates in which sulfotransferases transfer a sulfonate group from a donor molecule, i.e. 3’-
phosphoadenosine 5’-phosphosulfate (PAPS) to substrates (4, 200-201). Sulfonation is generally
a detoxification pathway because, as with other conjugation pathways, the products are more
water-soluble and thus should be more easily eliminated from the body (202). However, in some
cases, sulfonation can lead to bioactivation of compounds, leading to toxic products. For
example, sulfonation of N-hydroxyarylamines, N-hydroxy-heterocyclic amines, and
hydroxymethyl polycyclic aromatic hydrocarbons leads to reactive electrophiles, e.g. carbocation
or nitrenium ion intermediates, which are both carcinogenic and mutagenic (4). This is because
SO42-
is a good leaving group as discussed previously, which leads to reactive cations (Figure 4)
that can covalently bind to proteins and DNA (4).
Sulfation of minoxidil in rat skin is mediated by sulfotransferases (188). Rat skin also has ability
to synthesize PAPS, which is the cofactor for the sulfotransferases (203). Sulfotransferase-
mediated sulfation of another phenol, i.e. acetaminophen, was also shown in rat skin (204).
42
Figure 4. Reactive cations (electrophiles) formed by the loss of SO42- (adopted from (4)).
Drug bioactivation and covalent binding risk assessment have been a focus in drug development
in the pharmaceutical industry. As mentioned before, at Merck & Co., Inc., the quantity of
covalent binding to proteins has been used to guide drug development. If a drug candidate is
found to covalently bind to protein, especially if the binding is more than 50 pmole/mg protein,
the basis for that binding was investigated, and new analogs without the structural feature
responsible for the binding would be synthesized and tested until a structure is found that has
minimal binding and is likely to be safer (177). However, when corrected for daily dose,
covalent binding is related to the risk of liver toxicity, but it is not a perfect predictor of IDRs
(183). There are some drugs such as ximelagatran that do not appear to form reactive metabolites
and yet are associated with an unacceptable risk of liver toxicity that appears to be immune-
mediated (153).
43
3 Animal models
Animal models represent a major tool for mechanistic studies in virtually all of biomedical
research (205). Hypersensitivity reactions presumably involve a complex combination of genetic
and environmental factors as well as complex interactions between drug or metabolites and the
immune system that lead to their unpredictable nature, and a simple in vitro system is very
unlikely to be able to mimic such complexity. Although animals do have hypersensitivity
reactions to drugs and other xenobiotics, they are just as idiosyncratic in animals as they are in
people, so finding suitable animal models is very difficult and most attempts have failed. For an
animal model to be useful it should involve basically the same mechanism as the hypersensitivity
reaction in humans. The term ‘animal models’ that is used in our lab refers to a reaction in
animals in response to a drug that mimics the reaction occurring in some humans (1). One
interesting model is drug-induced anaphylaxis in mice. Mice were sensitized with either
penicillin V or cephalothin conjugated to ovalbumin via intraperitoneal injection. Two weeks
later, anaphylaxis was induced when an antibiotic-bovine serum albumin (BSA) conjugate was
administered intravenously (206).
Some drug-induced IDR animal models that have been tried but were not successful in producing
significant toxicity are the following:
Halothane causes idiosyncratic liver toxicity in humans. This toxicity is associated with
antibodies against trifluoroacetylated-proteins as well as auto-antibodies, which suggests that it is
immune-mediated. There have been many attempts to develop an animal model in guinea-pigs,
and mice. Halothane exposure resulted in an immune response in guinea pigs in which a single
exposure induced antibodies against trifluoroacetylated protein; however, it did not induce
44
significant liver toxicity in guinea pigs even after repeated halothane treatment, which only
induced elevated transaminases (207). In another experiment, Furst et al. also demonstrated a
cellular immune response (T cell activation) to trifluoroacetylated protein, which decreased with
additional exposure (208). In mice, acute liver toxicity was induced by halothane exposure;
however, delayed, immune-mediated liver failure was not successfully induced (209).
Aminopyrine causes agranulocytosis in humans and attempts were made to produce an animal
model in rabbits. It induced agranulocytosis in some rabbits and significantly depressed
granulocyte counts in other rabbits (210). However, we were not able to reproduce these results.
We also tried to induce agranulocytosis with clozapine and the analog DMP406, in mice, rats,
guinea pigs, and rabbits; however, clozapine does cause an increase in the rate of neutrophil
turnover in both rabbits and rats (211).
Amodiaquine was withdrawn from the market due to agranulocytosis and hepatotoxicity (180,
212). Anti-amodiaquine antibodies have been detected in humans with amodiaquine-induced
adverse reactions, indicating immune-mediated properties in these adverse reactions (213-214).
This is likely due to a reactive iminoquinone metabolite (215). In rats, amodiaquine treatment
also induced anti-amodiaquine antibodies, and at a dose of 538 mol/kg/day, it caused a
significantly reduced peripheral white blood cell count; however, a differential count was not
performed so it is not clear whether it caused neutropenia. The white blood cell count recovered
within a few days after amodiaquine was stopped. The ALT was also significantly increased
after amodiaquine was stopped and recovered two weeks later (216). The recovery of both white
blood cell count and ALT may not be mediated by immune tolerance, because the toxicity was
acute and recovery occurred after the drug was stopped. It is only when recovery occurs despite
continued treatment that it is reasonable to speculate that there is immune tolerance.
45
Felbamate is another drug that can cause both hepatotoxicity and aplastic anemia (217). The
reactive metabolite, atropaldehyde, formed spontaneously from the aldehyde carbamate
metabolite of felbamate, was believed to be responsible for these reactions (218). It was reported
that atropaldehyde constituted 1% and 6% of felbamate metabolites in rats and humans,
respectively (219). However, when rodents were treated with felbamate no hepatotoxicity or
hematotoxicity was observed even when protective pathways such as aldehyde dehydrogenase,
P450, glutathione S-transferase, and glucuronosyl transferase were inhibited (1). Procainamide
and hydralazine induced a lupus-like syndrome in humans, but treatment of mice with these
drugs did not lead to an autoimmune syndrome.
Even though most attempts to develop animal models have failed, there are a few successful
models that were discovered by accident. These models are penicillamine-induced autoimmunity
in Brown Norway rats, sulfonamide-induced hypersensitivity in dogs, propylthiouracil-induced
autoimmunity in cats, and NVP-induced skin rash in rats.
3.1 Penicillamine-induced autoimmunity in rats
Penicillamine is used in the treatment of Wilson’s disease, which is an autosomal recessive
genetic disorder. Manifestations of Wilson’s disease are neurological or psychiatric symptoms
and liver disease due to copper accumulation, and penicillamine treatment removes the excess
copper from the body. Penicillamine has also shown efficacy in the treatment of rheumatoid
arthritis, but it mediated autoimmunity (205). It causes a broad range of autoimmune reactions
including a lupus-like syndrome, pemphigus, and myasthenia gravis. It is also associated with
rash and agranulocytosis, which are likely immune-mediated (205).
46
Penicillamine-induced autoimmune disorders including drug-induced lupus were successfully
induced in rats, which was first described by Donker, et al. (220). Penicillamine treatment at 20
to 50 mg/day for 3-4 weeks induced weight loss, dermatitis, and circulating antinuclear
antibodies in Brown Norway rats, suggesting autoimmune involvement in this model. The
incidence was 73%. In another study, Tournade et al. found that penicillamine treatment
increased serum IgE levels and the number of CD4+ T cells and B cells in the spleen in Brown
Norway rats, indicating immune involvement in this model (221). The symptoms in this Brown
Norway rat model are similar to penicillamine-induced lupus in humans; therefore,
penicillamine-induced autoimmunity in Brown Norway rats appears to be a good animal model
to study drug-induced lupus in humans (1).
Penicillamine-induced autoimmunity appears to be specific to Brown Norway rats and does not
occur in Lewis or Sprague-Dawley rats (222). The dose-response curve is unusual: the incidence
with a 20 mg/day dose is between 50% and 80%, but the incidence is not increased by increasing
the dose to 50 mg/day. When Brown Norway rats received an escalating dose regimen (started at
5 mg/day, followed by 20 mg/day, and finally 50 mg/day by week 28), none of the rats
developed clinically-evident autoimmunity (220), which was confirmed in our lab. When rats
were treated at a dose of 5 to 10 mg/day, the incidence was 0%, and in fact, the lower dose
induced tolerance to the 20 mg/day dose (222). This is clearly immune tolerance as it can be
transferred to naïve animals with spleen cells or T cells from a tolerized animal (222-223). It was
found that CD4 T cells from tolerized rats treated with high-dose penicillamine (20 mg/day) had
increased levels of interleukin-10 (IL-10) and transforming growth factor (TGF)-ß mRNA, but
such elevations were not observed prior to high-dose penicillamine or in naïve animals treated
with high-dose penicillamine (222). These data suggest that the immune tolerance induced by
low-dose treatment is mediated by CD4+, CD25+ regulatory T cells, but the mechanism is likely
47
to involve additional cell types. Another study in our lab showed that the tolerance to
penicillamine requires both antigen presenting cells and T cells (223).
Although the incidence of penicillamine-induced autoimmunity is not increased by increasing
the dose beyond 20 mg/day, the incidence and severity are increased by poly-IC: a polymer of
inosine and cytosine that stimulates antigen-presenting cells via toll-like receptor 3 (224). Only
one dose of poly-IC given on the first day of penicillamine treatment is required even though
poly-IC has a short half-life, and on average it takes 3 weeks of penicillamine treatment before
the autoimmune syndrome becomes clinically apparent. Poly-IC treatment is also capable of
reversing tolerance; however, it does not significantly shift the penicillamine dose-response
curve: the combination of penicillamine at 10 mg/day and poly-IC does not lead to
autoimmunity. Furthermore, Lewis rats remain resistant: the combination of penicillamine at a
dose of 20 mg/d plus poly-IC does not cause autoimmunity in Lewis rats (224). These results
indicate that genetic factors are crucial for the susceptibility to penicillamine-induced
autoimmunity. Lipopolysaccaride, which stimulates macrophages through toll-like receptor 4,
had effects that were similar to those of poly-IC, but they were less pronounced (222). When the
rats became ill, they experienced significant weight loss and an increase in spleen weight, and all
treated rats present with increased serum IgE levels (222). Pretreatment of Brown Norway rats
with aminoguanidine, an inducible nitric oxide synthase inhibitor, and misoprostol, a
prostaglandin E analog, completely prevented the development of penicillamine-induced
autoimmunity (224). Increased splenic B7+ macrophages correlated with the incidence of
autoimmune disease, and the T cell inhibitor, tacrolimus, prevented disease onset, reversed
ongoing disease, and prevented disease relapse upon rechallenge with penicillamine. This
suggests that macrophages and T cells play important roles in the pathogenesis of this
autoimmune syndrome (225). Further investigation in our lab showed that covalent binding
48
between penicillamine and a macrophage surface aldehyde led to activation of macrophages. In
addition, there was a marked increase in Th17 cells, but only in animals that developed
autoimmunity (111-112, 226).
3.2 Sulfonamides in dogs
Sulfonamides are aromatic amines and are associated with a wide range of IDRs including a
generalized drug hypersensitivity reaction, skin rashes (including toxic epidermal necrolysis),
agranulocytosis, liver toxicity, and a lupus-like syndrome (205). The manifestations of
hypersensitivity usually include fever, skin rash, and involvement of other organs (227). The
incidence of sulfonamide-induced hypersensitivity was reported to be < 3% in HIV-negative
patients, but it increased to 65% in HIV-positive patients (228).
Sulfonamides also cause a range of drug hypersensitivity reactions in dogs, which are similar to
those that occur in humans and include fever, arthropathy, skin eruptions, thrombocytopenia,
hemolytic anemia, neutropenia, liver toxicity, etc. (205). Although sulfonamide-induced
autoimmunity can affect joints in humans, it is less common than in dogs. Larger breeds,
especially Dobermans, appear to be at higher risk than small breeds. One likely risk factor for
dogs is their inability to acetylate aromatic amines (205). Although the drug hypersensitivity
reactions induced by sulfonamides in dogs appear very similar to the reactions that occur in
humans making it a very attractive mechanistic model, the incidence is only ~0.25%. In addition
to the low incidence, it is difficult to work with dogs and so this is not a practical model (1).
3.3 Propylthiouracil-induced lupus in cats
Propylthiouracil is used for the treatment of hyperthyroidism, but its use is associated with
49
idiosyncratic liver toxicity, agranulocytosis, and a lupus-like syndrome (229). It also causes a
lupus-like syndrome in cats (230, Aucoin, 1988 #11258), which is characterized by lethargy,
fever, weight loss, antinuclear antibodies, and antimyeloperoxidase antibodies (231).
Propylthiouracil is oxidized to a reactive metabolite by myeloperoxidase (232), and this appears
to be a characteristic of several drugs that cause a lupus-like syndrome (233-234). Like
penicillamine-induced autoimmunity in the Brown Norway rat, rechallenge did not lead to a
shortened time to onset, although a second rechallenge did lead to a more severe reaction. It
remains to be determined whether lack of immune memory is a common feature of drug-induced
autoimmune reactions. It is not clear how important genetic determinants are in this syndrome
because mongrel cats were used and the incidence was ~40%. When our lab tried to reproduce
the syndrome at a later time we were unsuccessful; the only known change was a significant
increase in the level of taurine in cat chow because it was found that taurine deficiency in cats
leads to cardiomyopathy and other health problems (231). However, follow-up experiments to
determine if taurine deficiency is a risk factor for propylthiouracil-induced autoimmunity were
not performed (205).
3.4 NVP-induced skin rash model in rats
As mentioned before, the incidence of NVP-induced skin rash in humans is approximately 16%,
with 33% of those rashes being severe or life-threatening (20). Although more recent data from
Boehringer-Ingelheim indicated that the rate of NVP-attributable rash was reduced to 8.6%, with
20% of those rashes being severe or life-threatening (235), skin rash is a significant problem that
has restricted its use.
50
In the late 1990s, our lab attempted to establish an animal model to study the mechanism of
NVP-induced skin rash. This was based on a chance observation during a study of NVP
metabolism that 2 out of 4 NVP-treated female Sprague–Dawley rats developed erythema at 4-6
weeks. Other symptoms included excessive scratching around the nose/mouth area and loss of
body weight (236). Further investigation was performed on different rat strains/sexes, e.g. female
Brown Norway and Sprague–Dawley rats and male Brown Norway and Sprague–Dawley rats. In
2003, our lab had the good fortune to find that female Brown Norway rats developed a rash when
treated with NVP with an incidence of 100% (235).
The common features between the skin rashes in humans and female Brown Norway rats, which
are summarized in Table 1, indicated that NVP-induced skin rash in female Brown Norway rats
is very similar to that in humans, and therefore this is likely a good model to study the
mechanism of NVP-induced skin rash in humans (235). Although there was no liver toxicity in
this model, it represents those patients who develop skin rash only (235).
51
Table 1. A comparison of characteristics of NVP-induced skin rash in humans and female
Brown Norway rats (adapted from (1)).
Humans Rats
Rash Mild erythematous maculopapular rash to blistering skin eruptions
Mild to severe rash, no blisters
Time to onset 1-3 weeks after initiation of NVP 2-3 weeks after initiation of NVP
Dose response Incidence increases with dose Incidence increases with dose
Sex association Females are more susceptible Females are more susceptible
Tolerance Low dose treatment (200 mg/day) for 2 weeks significantly decreased incidence
Low dose treatment (40-75 mg/kg/day) for 2 weeks prevented skin rash
Rechallenge Rapid onset of skin rash and increased severity
Rapid onset of skin rash and increased severity
CD4 T cells Low CD4 T cell counts decreased incidence
Partial depletion of CD4 T cells decreased incidence
We have done extensive studies on NVP metabolism in this animal model. Metabolic pathways
of NVP in human and rats are outlined in Figure 5, which is adopted from Jie Chen’s paper (5)
and is based on an earlier study in mice, rats, rabbits, dogs, monkeys, and chimpanzees (237). In
that study, only a small fraction of the parent drug was excreted in urine (<6% of total urinary
radioactivity) and in feces (<5.1% of total fecal radioactivity) in all the species, while 41-46% of
total urinary radioactivity was excreted as parent drug in dogs. Hydroxylation, glucuronide
conjugation, and excretion in urine and feces were the major biotransformation and elimination
routes. The major hydroxylated metabolites were 2-, 3-, 8-, and 12-OH-NVP, and the other
52
major metabolite was 4-carboxy-NVP, which is formed by the further oxidation of 12-OH-NVP.
In rat plasma, the major species were NVP and 12-OH-NVP.
Figure 5. Major metabolic pathways of NVP (adopted from (5)).
In humans, the major routes of metabolism are also P450-mediated oxidation and
glucuronidation of the hydroxylated products (14). In one study, subjects took 200 mg NVP
tablets, once daily, for 2 weeks followed by 200 mg twice daily for 2 weeks, and glucuronidated
metabolites were found to represent the major metabolites in urine, i.e. 2-OH-NVP glucuronide
(18.6%), 3-OH-NVP glucuronide (25.7%), 12-OH-NVP glucuronide (23.7%), and 8-OH-NVP
glucuronide (1.3%). Hydroxylated metabolites in urine were 3-OH-NVP (1.2%), 12-OH-NVP
(0.6%), and 4-COOH-NVP (2.4%). Only 2.7% of parent drug was excreted in urine.
An in vivo study of reactive metabolites of NVP in patients tried to trap stable thioether
conjugates (6) to define the metabolism of NVP in patients and rats (Figure 6.). In patients’
urine, two isomeric NVP mercapturates were identified, which were also found in rat bile and
53
urine. NMR identified thioethers substituted at the C-3 (presumably from epoxide intermediates)
and exocyclic C-12 (from a quinone methide intermediate) positions of the methylpyrido ring of
NVP, suggesting that NVP undergoes bioactivation to arene oxide and quinone methide
intermediates. NVP-3-mercapturate was the major conjugate in urine, while NVP-12-
mercapturate was minor.
Figure 6. A proposed scheme of bioactivation and possible reactive metabolites of NVP (adapted
from (6)).
In rats, NVP can potentially form different reactive metabolites that are summarized in Figure 7
(adopted from the paper of Chen (5)). The hydroxyl groups on the 2- and 3- positions are para to
a nitrogen, and further oxidation could lead to quinoneimine type reactive metabolites. 12-OH-
NVP has the potential to be sulfated followed by loss of sulfate to form a reactive quinone
methide.
54
Figure 7. Putative bioactivation pathways of NVP (adapted from (5)).
In addition, 12-OH-NVP is further metabolized to 4-COOH-NVP, which when conjugated with a
good leaving group such as glucuronide or coenzyme A, has the potential to bind to protein (5).
The cyclopropyl structure of NVP can also form a free radical via one-electron oxidation by a
peroxidase (238). In the skin, peroxidases, such as prostaglandin synthase might oxidize the
cyclopropyl group and lead to opening of the ring with formation of a carbon-centered free
radical, which is more reactive than nitrogen-centered free radicals. In addition, although NVP
itself is not chemically reactive, it might bind directly to the MHC/TCR complex in a reversible
manner and induce an immune response (5).
55
At first our studies of NVP-induced skin rash in female Brown Norway rats demonstrated a good
correlation between NVP blood levels and the incidence of skin rash (5). They also showed that
increased blood levels of NVP when P450 was inhibited with aminobenzotriazole were
associated with an increased incidence of skin rash (5). It seemed that the concentration of
parent drug was critical for induction of the skin rash; however, evidence from further
investigations showed that one metabolic pathway, 12-hydroxylation, was responsible for NVP-
induced skin rash in rats as described in the following paragraphs.
NVP and 12-OH-NVP were the major species in plasma when rats were treated with NVP (150
mg/kg/day) in food with a peak NVP concentration of about 40 µg/mL after 7-8 days of
treatment (5). Then the NVP concentration decreased, which is consistent with the finding that
NVP induces P450s (11). In urine, the major metabolites were 2-, 3-, and 12-OH-NVP, and 4-
COOH-NVP (5). When P450 was inhibited by aminobenzotriazole, the excretion of 2- and 3-
OH-NVP and 4-COOH-NVP in urine were greatly decreased; however, 12-OH-NVP was not
significantly decreased. The reason for the lack of a decrease in 12-OH-NVP and a marked
decrease of 4-COOH-NVP is likely due to the role of P450 in both forming 12-OH-NVP and its
further oxidation to 4-COOH-NVP.
When lower doses of 12-OH-NVP (50 or 75 mg/kg/day) were administered to rats
subcutaneously, the incidence of skin rash was 100%, while NVP dosed at 75 mg/kg/day induced
a lower incidence (75%) of skin rash (5). This suggests that the rash is due to 12-OH-NVP rather
than NVP. To further test this hypothesis, an analogue of NVP in which the methyl hydrogens
were replaced by deuterium (DNVP) was synthesized. There should be less oxidation of the
methyl group in this analog because of the deuterium isotope effect, but all other properties of
56
the molecule should be virtually identical (Figure 8). Treatment of animals with this analog did
not lead to a rash which appeared to confirm the hypothesis (5).
Figure 8. Three major oxidative metabolites of NVP: 2-OH-NVP, 3-OH-NVP and 12-OH-NVP.
Replacement of the methyl hydrogens with deuterium (DNVP) decreases the formation of 12-
OH-NVP.
Since replacement of hydrogen with deuterium inhibits one of the major metabolic pathways of
NVP, we expected DNVP to have higher serum concentrations than NVP (5). However, the
serum concentrations of DNVP were markedly lower than those of NVP at the same dose. One
possible explanation is that the carbon free radical formed as an intermediate from NVP
oxidation by P450s could partition between oxygen rebound to form 12-OH-NVP or loss of
another hydrogen atom to form the reactive quinone methide (Figure 9). The quinone methide is
57
Figure 9. A putative bioactivation pathway of NVP in the liver. NVP is oxidized by cytochromes
P450 to a free radical intermediate that can partition between oxygen rebound to produce the 12-
OH-NVP and loss of another hydrogen atom to directly produce the quinine methide.
reactive and would likely bind to P450s to inhibit them, which would lead to higher NVP
concentrations. DNVP would form less reactive quinone methide resulting in less P450
inhibition and lower DNVP concentrations (5). To counter the decreased inhibition by DNVP,
animals were co-treated with aminobenzotriazole, which led to similar concentrations of NVP
and DNVP, but the concentration of 12-OH-NVP was lower in the DNVP-treated rats, and the
incidence of rash was also decreased to 20% and the rash was milder (5). This provides very
strong evidence that the 12-hydroxylation pathway is required to induce the NVP skin rash. 12-
OH-NVP is not chemically reactive, but it is a benzylic alcohol, and it could be converted to the
quinone methide, not by oxidation because the alcohol and quinone methide are the same
oxidation state, but by forming a conjugate that adds a better leaving group than hydroxide. The
58
only reasonable candidate is the sulfate, and as mentioned earlier, there are sulfotransferases in
the skin (Figure 10) (181).
Figure 10. A putative bioactivation pathway of NVP in the skin. NVP is oxidized by
cytochromes P450 to 12-OH-NVP in the liver. Sulfotransferases in the skin transfer a sulfate
group to 12-OH-NVP to form a sulfate conjugate. Attack by nucleophiles including proteins
would lead to covalent binding.
We speculated that the reactive quinone methide may be responsible for both liver toxicity and
skin rash in humans (184). However, the sulfate is not as reactive as expected, and it does not
directly form the quinone methide, but it does appear to bind to skin proteins by an SN2
mechanism (Figure 10).
59
One in vitro study of NVP oxidation with human liver microsomes and GSH found that CYP3A4
was the primary enzyme leading to a GSH conjugate, the structure of which suggests that it came
from the quinine methide (239). Other in vitro studies using a synthesized electrophilic 12-
mesyloxy-NVP found that it formed DNA, amino acid, and protein adducts (240-242); however,
mesolate is a much better leaving group than sulfate, and it is not formed in animals or humans
so these experiments are irrelevant.
When female Brown Norway rats were treated with NVP at a dose of 150 mg/kg/day, the first
sign of a reaction was ear redness after 7-10 days of treatment (235). Rash with scabbing on the
back usually appeared at about 21 days of treatment. Histology of the skin showed mononuclear
infiltration, and immunohistochemistry demonstrated that the infiltrate was composed of CD4
and CD8 T cells and macrophages (235). When sensitized animals were rechallenged with NVP
after they recovered (about two weeks off drug), ear redness appeared within 24 h. At
rechallenge, skin rash was less obvious that on primary treatment (235), but it occurred earlier
(after about 9 days) and the histology showed a more extensive infiltrate, and unlike the animals
on primary exposure, the animals appeared sick (243). Interestingly, splenocytes (T cells, most
likely CD4 T cells) from rechallenged animals were able to transfer susceptibility to NVP-
induced skin rash to naive female Brown Norway recipients; specifically, the recipients
developed red ears in about 8 h and became ill much like rechallenged animals (244).
Rechallenge also led to an increase in the total number of auricular lymph node T and B cells as
well as macrophages in which the activation/infiltration marker, intracellular adhesion molecule-
1 (ICAM-1) and activation/antigen presentation marker MHC II increased as well (243). In the
ears, primary treatment with NVP led to macrophage infiltration and ICAM-1 expression as early
as day 7 of treatment, but T cell infiltration was not apparent until the onset of rash. In addition,
when the rash developed, both MHC I and MHC II expression was increased (243).
60
In both the penicillamine-induced autoimmunity and NVP-induced skin rash models, tacrolimus
prevented the adverse reaction, which is consistent with an immune mechanism. In addition, low
dose treatment with the drug for 2 weeks prevented the adverse reaction induced by treatment
with a full dose of the drug. In the penicillamine model the mechanism of protection is immune
tolerance because it can be transferred to naïve animals with spleen cells, but in the NVP model
it appears that the mechanism of tolerance induced by low dose treatment is mostly due to
induction of P450 because it can be prevented by the P450 inhibitor, aminobenzotriazole, and it
is not long lasting like the tolerance induced by low dose treatment in the penicillamine model
(235).
In humans, higher CD4 T cell counts were associated with a higher incidence of skin rash in
NVP treated-patients (24). When sensitized patients were rechallenged with NVP, the onset of
skin rash was faster (22), suggesting immune memory. As mentioned earlier, the low dose drug
regimen (200 mg/day) for two weeks can partially protect against the skin rash when patients
were later put on the regular dose 200 mg twice a day (20), suggesting tolerance induction.
3.5 Danger signals in NVP-induced skin rash
3.5.1 Danger signals in IDRs
Our lab has studied potential danger signals, as determined by changes in mRNA expression, in
other animal models of IDRs. One example was the analysis of mRNA changes in the liver of
penicillamine-treated male Brown Norway rats (245). Gene expression 6 h after dosing showed
61
changes in genes that have a role in stress, energy metabolism, acute phase response, and
inflammation (245).
Tienilic acid was withdrawn due to idiosyncratic hepatotoxicity, and it was reported that human
CYP 2C9 and rat CYP 2C11 metabolize tienilic acid to a reactive thiophene epoxide that reacted
selectively with the P-450 that formed it (246-247). P450 is not an essential enzyme for cell
survival, and therefore it seemed likely that this binding would not lead to cell stress. This could
represent a test of the danger hypothesis, but the experiment found changes in expression of
genes involved in oxidative stress (aldo-keto reductase, glutathione-S-transferase, thioredoxin
reductase, epoxide hydrolase), inflammation (IL-1β, interferon regulatory factor 1, macrophage
stimulating protein 1), cytotoxicity (caspase-12), and liver regeneration (p27Kip1, DUSP6,
serine dehydratase, spectrinβII, inhibin βA) at 6 and 24 h after drug administration in rats (248).
These results are consistent with the danger hypothesis, and it was later found that the reactive
metabolite of tienilic acid bound to several proteins, not just P450 (249).
The aromatic anticonvulsants carbamazepine and phenytoin are also associated with a relatively
high incidence of idiosyncratic drug reactions (IDRs), which appear to be immune-mediated. We
also found that major metabolites of these two drugs: 3-OH-CBZ and 4-OH-PHN can be
oxidized by peroxidases to phenoxyl free radicals, which could cause oxidative stress by redox
cycling (250). Microarray analysis showed that CBZ and PHN treatment induced changes in
mRNA expression of the liver in mice, many of which were related to Keap1-Nrf2-ARE
signaling pathways and enzymes involved in responding to oxidant stressors and reactive
metabolites such as glutathione transferase and HSPs (251). These gene changes, which
represent danger signals, were most likely due to cell stress induced by reactive metabolites of
CBZ and PHN (251).
62
Sulfamethoxazole is an aromatic amine, and aromatic amines in general are associated with a
relatively high incidence of idiosyncratic drug reactions. This is presumably because they are
oxidized to nitroso electrophiles, and in addition, they can redox cycle, Therefore, we expected
that sulfamethoxazole would cause many changes in gene expression in the liver; however, no
gene changes in the liver that can be interpreted as a danger signal were induced by
sulfamethoxazole (248). In retrospect, although sulfonamide-induced hypersensitivity reactions
can involve the liver, most sulfonamide-induced IDRs are in fact cutaneous. Most of the redox
cycling may occur outside the liver and the liver is better equipped to detoxify reactive
metabolites than other organs.
63
4 Hypothesis
Danger signals released by skin cells initiate the immune response responsible for NVP-induced
skin rash
4.1 Strategy
NVP forms reactive metabolites that covalently bind to proteins, and this could induce cell stress
and release danger signals. In addition, NVP-induced skin rash has been shown to be immune-
mediated; therefore, these reactive metabolites may produce danger signals that are involved in
the initiation of NVP-induced skin rash. Given that the skin is the target of a skin rash, it is
logical to study potential danger signals induced by NVP in the skin. Although the time to onset
of skin rash induced by NVP is red ears at 7 days and skin rash at 21 days of primary treatment,
it takes time to mount an adaptive immune response, and it is likely that the cell injury/danger
signal occurs very early. We also want to detect early gene changes and avoid secondary or
downstream effects of the primary response; therefore, 6 are 12 h were chosen for sample
collection. The first visible change is in the skin of the ear and so the first study used the ear for
analysis, and although under the right conditions other rat strains can develop NVP-induced skin
rash, the most reliable is the female Brown Norway rat. As describe earlier, we know that 12-
OH-NVP is required for induction of the rash; therefore, the effects of 12-OH-NVP were
compared with those of NVP. We also know that substitution of methyl hydrogen atoms with
deuterium decreases the conversion of NVP to 12-OH-NVP (5); therefore, the effects of DNVP
were compared with those of NVP.
64
As mentioned above, NVP metabolism in the liver catalyzed by cytochromes P450 directly
produces the reactive quinone methide, and although no liver toxicity was observed in rats
treated with NVP, it seemed likely that some danger signals would be produced and so changes
in gene expression in the liver were also determined. In order for danger signals to induce an
immune response they must be able to stimulate cells of the immune system, presumably antigen
presenting cells, and therefore the focus was on changes in the expression of molecules that are
released from cells or expressed on their surface.
65
5 Materials and Methods
5.1 Materials
NVP and ethyl-NVP (a NVP derivative in which the cyclopropyl group was replaced by an ethyl
group) were kindly supplied by Boehringer-Ingelheim Pharmaceuticals, Inc., Ridgefield, CT.
PBS (without calcium and magnesium, 150 mM, pH 7.4) was obtained from the University of
Toronto Media Services (Toronto, ON). Rabbit polyclonal anti-HMGB1 was obtained from
Abcam (Cambridge, MA), and a HMGB1 ELISA kit was purchased from IBL International
GMBH (Hamburg, Germany). Protease inhibitor cocktail and anti-rabbit IgG peroxidase
conjugate produced in goat were purchased from Sigma (Oakville, ON). All primers were
ordered from Integrated DNA Technologies (Coralville, Iowa) in standard desalting medium.
Immunohistochemistry reagents including biotinylated anti-rabbit IgG (H+L) produced in goat,
normal goat serum, avidin/biotin blocking kit, horseradish peroxidase avidin D, 3,3'-
diaminobenzidine substrate kit for peroxidase and hematoxylin were from Vector Laboratories,
Inc, (Burlingame, CA). Omniscript reverse transcription kit, RNase-free DNase set, and
RNAlater RNA stabilization reagent and RNeasy mini kit were from Qiagen (Hilden, Germany).
Protector RNase inhibitor, primer p(dT)15 for cDNA synthesis and Lightcycler Faststart DNA
Master SYBR green was from Roche (Mannheim, Germany). Anti-Nr4a3 antibody was obtained
from Santa Cruz Biotechnology, Inc. (Santa Cruz, Ca). In-gel tryptic digestion kit was from
Pierce (Rockford, IL). Cell lysis buffer was from Cell Signaling Technology (Danvers MA).
Anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was from LabFrontier (Seoul,
Korea). PVDF membrane, pure nitrocellulose membrane (0.2 µm), protein Bradford assay,
ReadyStrip (immobilized PH gradient gel strip, IPG strip) of pH range 3-10 and β-
66
mercaptoethanol were purchased from Bio-Rad Laboratories (Mississauga, ON).
Paraformaldehyde (16%) was obtained from Canemco & Marivac Supplies (Quebec, Canada).
Amersham ECL plus western blotting detection reagents was from GE Healthcare
(Buckinghamshire, UK). All solvents used for HPLC and mass spectrometry were HPLC grade.
IL-22 receptor alpha 2 (IL-22ra2) ELISA kit was purchased from Life Science Inc. (Wuhan, The
P. R. China)
5.2 Methods
5.2.1 Animal Care
Female Brown Norway rats (150-175 g) were obtained from Charles River (Montreal, QC) and
housed in pairs with a 12:12 h light/dark cycle with free access to water and Agribrands pellet
lab chow (Leis Pet Distributing, Inc., Wellsley, ON). Animals were acclimatized for 1 week
before experiments were initiated. When animals were administered with drug in food, they were
put on powdered chow (rodent meal 2018, Leis Pet Distributing, Inc., Wellsley, ON) for one
week before drug treatment. At the end of the experiment, rats were killed by carbon dioxide
asphyxiation followed by cervical dislocation. All of the animal studies were conducted in
accordance with the guidelines of the Canadian Council on animal care.
5.2.2 Drug administration
NVP (150 mg/kg/day), 12-OH-NVP (159 mg/kg/day), or DNVP (151 mg/kg/day) were
administered to rats via i.p., gavage, or in food. All drugs for administration via i.p. or gavage,
were prepared as a suspension in 0.5% methylcellulose (MC, vehicle), and control rats were
either non-treated or treated with same amount of MC. When drugs were administered in food,
they were mixed with the powdered chow, and control animals were given plain powdered chow.
67
5.2.3 Synthesis of 12-OH-NVP
The synthesis of 12-OH-NVP followed the method described in Chen et al. (5) with some
modifications. To a flame-dried round bottomed flask equipped with a magnetic stirrer, 6.1 g of
NVP, oven-dried at 60 °C overnight, was added. The flask was sealed and equipped with a
nitrogen balloon. Anhydrous tetrahydrofuran was added, the solution was cooled to -78 °C and
140 mmol lithium diisopropylamide was added over a period of 5 min. The solution was kept at -
78 °C for 2 h with stirring. Then the reaction mixture was allowed to warm to -40 °C, and
anhydrous oxygen was bubbled through the solution over 4 h while the temperature was
maintained at -40 to -20 °C. The now clear solution was acidified with 2 N hydrochloric acid
over ice and the organic layer was extracted with 3 x 30 mL of 2N hydrochloric acid. The
combined aqueous layers were brought to pH 8 using sodium carbonate and extracted with
4x100 mL of methylene chloride. The combined organic layers were washed with brine and
water, dried over anhydrous magnesium sulfate and evaporated in vacuo to yield a solid product.
The crude product was purified using open column chromatography with silica gel (Sigma-
Aldrich, pore size 60 Å, 70 – 230 mesh, column dimensions 40 x 400 mm). The solvent system
used was hexanes: ethyl acetate starting at a proportion of 60:40 and increasing to 100% ethyl
acetate to yield a fluffy pale yellow powder in 20% yield. The amount of NVP contamination of
the obtained product was analyzed using mass spectrometry in the multiple reaction monitoring
mode (MRM). The ion pairs used for the analysis were: 267.0/226.1 for NVP and 283.1/223.1
for 12-OH-NVP.
68
5.2.4 Synthesis of NDVP
The synthesis of DNVP followed the method described in Chen et al. (5) with some
modifications. To a flame-dried flask was added NVP (1.0 g, 3.7 mmol), followed by potassium
tert-butoxide (0.8 g, 7.4 mmol) and DMSO-d6 (24 mL, 342.8 mmol), and the mixture was
refluxed at 140 °C under argon for 48 h. The reaction mixture was diluted with cold water (100
mL) and extracted with ethyl acetate (200 mL). The ethyl acetate layer was then washed with
brine (200 mL × 2), dried over anhydrous sodium sulfate, and concentrated to yield crude
product, which was column purified using ethyl acetate to yield 0.9 g of product as a yellow
solid in 97% yield. 1H NMR (CDCl3): δ 0.31-0.41 (m, 2H), 0.83-0.90 (m, 2H), 3.60-3.64 (m,
1H), 7.06 (d, J ) 4.8 Hz, 1H), 7.19 (dd, J ) 4.8, 7.5 Hz, 1H), 8.01 (dd, J ) 2.1, 6.6 Hz, 1H), 8.08 (d,
J) 4.8 Hz, 1H), 8.50 (dd, J) 1.8, 4.8 Hz, 1H), 9.90 (bs, 1H). ESI-MS 270 (MH+), ratio of the peak
heights of 267:268:269:270 was 0:0.007:0.124:0.869, which indicated that synthesized DNVP
contained only traces of NVP.
5.2.5 Mass spectrometry
Plasma (10 µL) was diluted with 10 µL water and mixed with internal standard solution (ethyl-
NVP in methanol, 5.4 µg/mL, 20 µL). A standard solution of NVP and 12-OH-NVP was
prepared ranging from 4.0 to 65 µg/mL. Each standard solution (10 µL) was mixed with control
plasma (10 µL), and then mixed with internal standard solution (20 µL). To each prepared
sample and standard, methanol (60 µL) was added and cooled to -20 oC for 30 min to precipitate
protein. After incubation, the samples were centrifuged at 16000Xg for 10 min, and then 20 µL
of supernatant from each sample was mixed with 180 µL mobile phase (20% acetonitrile and
80% water with 2 mM ammonium acetate and 1% acetic acid).
69
The samples were separated by HPLC and analyzed by mass spectrometry. The separation was
carried out on an Ultracarb C18 30 X 2.0 mm, 5 µm column (Phenomenex) under isocratic
conditions with a mobile phase consisting of 20% acetonitrile and 80% water with 2 mM
ammonium acetate and 1% acetic acid and the flow rate of 0.2 L/min. Mass spectrometry was
performed with a PE Sciex A 3000 quadrapole system and an electrospray ionizing source and
analyzed by Analyst Software. The data acquisition method was MRM for NVP and 12-OH
NVP serum level measurement. The ion pairs used for the analysis were: 267.0/226.1 for NVP,
283.1/161.0 for 2-OH-NVP, 283.1/214.0 for 3-OH-NVP, 283.1/223.1 for 12-OH-NVP,
297.1/210.1 for 4-COOH-NVP, 255.1/227.2 for ethyl-NVP (positive ionization mode),
361.0/96.0 for 12-sulfoxy-NVP and 229.0/169.8 for naproxen (negative ionization mode).
Standard curves prepared for 2-OH-NVP (0.43 – 102.9 µg/mL), 3-OH-NVP (0.36 – 86.8
µg/mL), 12-OH-NVP (0,38 – 91.0 µg/mL), 4-COOH-NVP (0.26 – 61.8 µg/mL), 12-sulfoxy-
NVP (0.28 – 14.0 µg/mL) and NVP (0.74 – 176.9 µg/mL) had R2 values of > 0.99.
5.2.6 Microarray study of rat liver, skin, whole ear and ear skin
For the microarray study of whole ear or peeled ear skin, NVP or 12-OH-NVP was administered
via i.p., and control rats were not administered either drug or vehicle. The number of rats for
each drug and time point is stated in Table 2-5. Samples of about 0.5 x 0.3 cm2, taken from
whole ear or peeled skin and cleared of hair and fat, were put in RNAlater RNA stabilization
reagent according to manufacturer’s instructions.
For the microarray study of liver, NVP (n=4 for each time point) or 12-OH-NVP (n=4 for each
time point) was administered via gavage and control rats (n=4 for each time point) were
administered via gavage the same amount of vehicle. Six or 12 h after treatment, the rats were
killed and blood was obtained by cardiac puncture to determine NVP and 12-OH-NVP plasma
70
levels. The size of liver sample was about 0.5 x 0.3 cm2, and the site of sampling was consistent
for each animal. The liver samples were put in RNAlater RNA stabilization reagent according to
manufacturer’s instructions.
For the microarray study of skin, NVP (n=4) or 12-OH-NVP (n=4) was administered in the same
way as in the above liver study. The only difference was that there was only one time point (6 h)
in the skin study. Samples of about 0.5 x 0.3 cm2, taken from back and cleared of hair and fat,
were put in RNAlater RNA stabilization reagent.
RNA extraction was performed on all the samples with RNeasy mini kits and RNase-free DNase
kits according to the manufacturer’s instructions. Extracted RNA samples were sent to the
Hospital for Sick Children Microarray Centre, Toronto for analysis.
Data analysis was performed with Partek genomics suite software to identify RNA expression
changes by comparing the treatment group with the control group. One-way or 2-way ANOVA
was used for the microarray data analysis. The statistical significance of changes in gene
expression was determined by the False Discovery Rate (FDR) <0.05 filter. Genes that pass the
filter were considered significant. Gene symbols were consistent with those of the NCBI gene
bank.
5.2.7 Immunohistochemistry
Ears were removed from the rats administered with NVP (i.p.) or vehicle (i.p.). and fixed with
4% paraformaldehyde. Cryosectioning was performed and ear sections were cut at a thickness of
15 µm and placed on slides. Immunohistochemistry was performed as follows: First, sections
were incubated with 1% H2O2 and 2% NaN3 for 60 min to block endogenous peroxides, which
was followed by rinsing with PBS solution for 5 min x 3. After washing, sections were blocked
71
in 1% normal horse serum diluted in PBS. Then sections were further blocked in avidin D
solution for 15 min and then blocked in biotin solution for another 15 min. Next, sections were
incubated with anti-HMGB1 (diluted in PBS at 1:200) for 1 h, which was followed by rinsing in
PBS solution for 5 min x 3. After rinsing, the sections were incubated with horse anti-mouse
biotinylated antibody 5 µg/mL (prepared in 1% horse serum) for another 1 h. Then sections were
incubated with peroxidase-conjugated avidin for 45 min in the dark, which was followed by
incubation with peroxidase substrate DAB for 5 min (prepared according manufacture’s
instruction). Sections were then washed with tap water and counter stained with hematoxylin for
30 seconds followed by mounting with glycerol and a cover glass.
5.2.8 Synthesis of rabbit anti-rat S100a7a antibody
The amino acid sequence (108aa) of rat S100a7a was obtained from the NCBI protein bank. The
C terminal 16 amino acid sequence was chosen to synthesize a peptide, which was conjugated
with KLH and injected into two rabbits to induce anti-rat S100a7a antibodies. After three booster
injections, titration of the anti-sera was done by western blotting. The antibody at a dilution of
1/1000 detected 1µg of the peptide.
5.2.9 Western blotting
The ear or skin tissue was homogenized with Ultra-Turrax T25 homogenizer (Janke & Kunkel,
Staufen, Germany) and then centrifuged at 16000 Xg at 4 °C for 15 min. The supernatant was
analyzed by SDS-PAGE electrophoresis with a Bio-Rad mini-Protean 3 cell. Protein (30 µg) was
loaded in each well on 10 -15% SDS-PAGE gels. A voltage of 80 volts for 20 min was used for
the stacking gel, while 150 volts was used for the separating gel. After proteins were separated,
they are transferred to a nitrocellulose membrane at 100 volts for 1 h. The transferred membranes
72
were blocked with 1% skim milk solution for 1 h at 4 oC and then incubated with primary
antibody solution (diluted in PBS according to each manufacture’s suggestion) for 1 h at 4 oC.
After washing in PBS plus Tween-20 (0.05%) solution for 5 min three times, the membranes
were incubated in HRP-conjugated secondary antibodies (dilution according to each
manufacture’s suggestion) for 1 h at 4 oC. After washing with PBS plus Tween-20 solution for 5
min three times, membranes were put in ECL plus reagent for 5 min incubation and then imaged
with a FluorChem 8800 imager (Alpha Innotech, CA). Western blotting was performed for
S100a7a and HMGB1 proteins in ear tissue lysate from administered with NVP (i.p.) and control
rats.
5.2.10 2D-electrophoresis
2D-electrophoresis was performed for HMGB1 protein in the ear in a NVP-treated (i.p.) rat and a
control rat according to the Bio-Rad 2D-electrophoresis protocol. The ears were homogenized
with an Ultra-Turrax T25 homogenizer in lysis buffer (20 mM Tris-HCL pH 6.8, 7 M urea and 2
M thiourea). The lysate was then centrifuged at 16000g at 4 °C for 15 min after which the
supernatant was taken and mixed with rehydration buffer (7 M urea, 2 M thiourea, 2% CHAPS,
1% DTT, 0.2% biolyte-ampholyte). During rehydration, a 7 cm dry, 2 mm in width IPG gel was
incubated in 125 µL of sample solution (about 100 µg protein) overnight.
For the isoelectric focusing (IEF) the loaded gel was put in a Bio-Rad Protean IEF cell and run at
8,000-10,000 volt-hour for about 2-3 h with a maximum voltage of 4000 V at 20 oC. Because the
isoelectric point (PI) of HMGB1 is 5.62, a pH 3-10 IPG gel was used. After IEF, the gel was put
in equilibration buffer (6 M urea, 2% SDS, 0.375 M Tris-HCL, pH 8.8, and 20% glycerol) to
reduce disulfide bonds and to alkylate the resultant sulfhydryl groups of the cysteine residues.
After equilibration, the gel was ready for second-dimension separation.
73
For the second-dimension separation the gel was put on top of a SDS-PAGE gel in a Bio-Rad
mini-Protean 3 cell. During this separation, proteins migrated from the IPG gel to the SDS-
PAGE gel and were separated on the basis of their molecular mass. Based on the molecular mass
of HMGB1 of 24 KD, a 10% SDS-PAGE gel was made in house.
After second-dimension separation was completed, proteins on the SDS-PAGE gel were
transferred to a nitrocellulose membrane, which was then incubated with primary anti-HMGB1
antibody for 1 h. After washing in PBS plus Tween-20 (0.05%) solution for 5 min three times,
the membrane was incubated in HRP-conjugated secondary antibody for 1 h at 4 oC. After
washing with PBS plus Tween-20 solution for 5 min three times, the membrane was put in ECL
plus reagent for 5 min and then imaged with a FluorChem 8800 imager to visualize HMGB1
protein on the membrane.
5.2.11 ELISA analysis
ELISA analysis for serum level of HMGB1 in rats treated with NVP (i.p.) or IL-22ra2 in rats
treated with NVP or 12-OH-NVP (in food) was performed according to the manufacture
instructions. The optical density reading was done at 450 nm. The correlation coefficient (non-
linear 4-parameter regression) of the standard curve for both proteins was > 0.99.
5.2.12 Real time-PCR
Real time-PCR was performed for genes of interest in rats after NVP, 12-OH-NVP or DNVP
treatment (i.p.) and control rats. The interesting genes were collected from the microarray study
with LightCycler 2.0 instrument (Roche) and DNA master SYBR green kit and Lightcycler
software-3.5 (version 5.32).
74
Experimental protocol: 40 cycles for each run; pre-incubation, 95 oC, 10 min; amplification, 95
oC, 10 min, 60 oC, 5 min, 72 oC, 10 min; temperature transition, 20 oC/sec; melting, 60 oC, 30
min. The data analysis software was RelQuant.
The standard curve for each gene was done by serial dilution: 1:1, 1:10, 1:100, 1:1000; 1:10000.
Efficiency=10 -1/slope. An acceptable slope of a standard curve was from -3.1 to -3.7.
The genes that were studied were Nr4a3, S100a7a, MT-1, MT-2, Fkbp5, and HMGB1.
Primers:
HMGB1: 5’-CCG GAT GCT TCT GTC AAC TT-3’ (Forward), 5’-TTG ATT TTTGGG CGG
TAC TC-3’ (Reverse);
NR4a3: 5’-TAT CCT TTG TTT GCA GTG ACC TTT A-3’ (Forward), 5’-TCT TCA AAC GTT
ATT TGA ATT TAG C-3’ (Reverse);
S100a7a: 5’-TCT GCA GAT TTG CCT GTA CCC TGA-3’ (Forward), 5’-TGA AGC GAG
GCA CAC TAT CCA AGA-3’ (Reverse);
Mt2: 5’-ACA GCG ATC TCT CGT TGA TCT CCA-3’ (Forward), 5’-GCA TTG TTT GCA
TTT GCA GGA GCC-3’ (Reverse);
Mt1a: 5’-ACC GTT GCT CCA GAT TCA CCA GAT-3’ (Forward), 5’-AGG AGG TGC ATT
TGC AGT TCT TGC-3’ (Reverse);
Fkbp5: 5’-CAC TTC TGC CTC CTT GCG TTG TTT-3’ (Forward), 5’-AGG GTC GCC CAA
GTT AGA ACA AGT-3’ (Reverse)
75
B2M: 5’-ATG GGA AGC CCA ACT TCC TCA ACT-3’ (Forward), 5’-TCT CGG TGG GTG
TGA ATT CAG TGT-3’ (Reverse)
76
6 Results
6.1 Microarray analysis of gene expression changes in the whole
ear tissue or peeled ear tissue after NVP, 12-OH NVP, or DNVP
treatment for 6 or 12 h
In general, no significant gene expression changes were identified in the whole ear tissue or
peeled ear tissue 6 or 12 h after NVP, 12-OH-NVP, or DNVP treatment. The first microarray
study was performed with samples from the whole ear 6 or 12 h after NVP treatment (Table 2).
Although there were apparent changes in gene expression, none of the changes reached statistical
significance. Therefore, they may be useful for hypothesis generation but no definitive
conclusions can be drawn. The gene symbols used in this table are consistent with the NCBI
gene bank nomenclature. Genes with possible biological significance included: Nr4a3, a nuclear
receptor that is an early response gene involved in many cellular functions, was only apparently
up-regulated at 6 h; Ddit4, DNA-damage inducible transcript 4, apparent up-regulation of 3 fold
at 6 h; FKBP5, an immunosuppressive drug-binding protein, apparent up-regulation of 2 fold at 6
h and 2 fold at 12 h; S100a3, a danger signal, apparent up-regulation of 3 fold at 12 h; S100A15
(now referred to as S100a7a), a danger signal, apparent up-regulation of 2 fold at 6 h and 2 fold
at 12 h.
More genes were apparently down regulated with high fold changes at 12 h than at 6 h.
77
Table 2. Genes with apparent high fold, but statistically non-significant, changes in whole rat ear
6 h (column A) or 12 h (column B) after NVP treatment. Numbers represent the apparent fold
change: positive = up-regulation, negative = down-regulation. Statistical significance was
determined by a one-way ANOVA analysis. Note: S100A15 is referred to later in the thesis as
S100a7a. (Rat 230 2.0 chips; NVP 6 h, n=2; NVP 12 h, n=2; controls, n=2)
The second microarray study was performed with samples from the whole ear 6 or 12 h after 12-
OH-NVP or 6 h after NVP treatment (Table 3). The 12-OH-NVP metabolite is known to cause
A B
78
the rash and the NVP treatment was added to try to allow comparison between this experiment
and the previous experiment with NVP. As with the previous experiment, none of the gene
changes reached statistical significance, but they may provide clues for hypothesis testing. Genes
of potential biological interest included: Nr4a3, a nuclear receptor, which is an early response
gene involved in many cellular functions, with an apparent up-regulation of 3 fold at 6 h and 3 at
12 h; FKBP5, a immunosuppressive drug binding protein with an apparent up-regulation of 3
fold at 6 h; IL-22ra2, a soluble antagonist of IL-22, with an apparent up-regulation of 4 fold at 12
h. Some genes, e.g. Nr4a3, Ddit4, and Snfilk, with high fold changes after NVP treatment in the
first experiment were found with high fold changes after NVP treatment in the second
experiment, indicating consistency and comparability between these two groups of microarray
studies.
79
Table 3. Genes with apparent high fold, but statistically non-significant, changes in whole ear 6 h
(column A) or 12 h (column B) after 12-OH-NVP treatment, or 6 h after NVP treatment (column
C). Statistical significance was determined by a one-way ANOVA analysis. (Rat 230 2.0 chips;
NVP 6 h, n=1; 12-OH-NVP 6 h, n=2, 12 h, n=2; controls, n=3)
Gene Fold Gene FoldNr4a3 3.13 Txlnb -2.52Fkbp5 3.09 Mlf1 -2.52Crebbp 3.04 Txlnb -2.54Fkbp5 2.95 Col10a1 -2.54Mt1a 2.92 Myl2 -2.54Tgfb2 2.33 Myh7 -2.55Hmgcs2 2.33 Ddit4l -2.56Ptgfr 2.20 Apobec2 -2.56Agt 2.12 Myh7 -2.59Trps1 2.09 Pvalb -2.62Mt2A 2.07 Rnase2 -2.64Pck1 2.07 Aqp4 -2.73Sf3b1 2.06 Mstn -2.74Slc7a5 2.05 Ky -3.30Fcgr2b 2.04 LOC680367 -4.53Cebpd 2.03RGD131086 2.01Peli2 2.01
Gene Fold Gene FoldCrebbp 3.40 Myh6 /// -2.00Tmed5 2.50 Mstn -2.00RGD156456 2.35 Myh7 -2.03Ptgfr 2.31 Myl2 -2.04Nr4a3 2.30 Myh7 -2.05Trps1 2.25 Trdn -2.08Atrx 2.20 Fos -2.12Ash1l 2.17 Pvalb -2.30Ddit4 2.12 Mylk2 -2.33Slc6a6 2.09 Mybpc2 -2.42Xiap 2.09 Actn3 -2.43Foxo1 2.07 LOC680367 -4.51Snf1lk 2.04Sp1 2.01
Gene Fold Gene FoldLOC497995 4.04 Pygm -4.03Mt4 4.00 Ckm -4.04Krt25 3.34 Tnnc2 -4.07Krt34 3.22 Myom2 -4.19Tchh 3.07 Actn3 -4.2S100a3 3.00 Eno3 -4.2Krt31 2.96 Neb -4.23Mt1a 2.78 Kbtbd10 -4.37RGD156246 2.71 Casq1 -4.43Mt1a 2.70 LOC688915 -4.53LOC682990 2.53 Hfe2 -4.54Fkbp5 2.40 LOC684425 -4.69Tcfcp2l1 2.29 Fos -4.69Krt33a/b 2.27 Myh1 -4.83Krtap14l 2.25 Trdn -5.19LOC683613 2.21 Ppp1r3a -5.33Mt2A 2.20 Pvalb -5.72LOC688990 2.15 Ky -6.48Gjb2 2.14 Myh4 -8.03Klf15 2.12S100a7a 2.11Slc28a2 2.10Hmgcs2 2.09Slc7a5 2.09Klf15 2.08Krt86 2.04Angptl4 2.01
The ear contains a lot of connective tissue including cartilage that is unlikely to respond to the
drug and this may dilute any changes that may have occurred in the skin, especially the thin
epidermis. Therefore, an experiment was performed with samples from skin peeled from the ear
6 h after NVP treatment (Table 4). Again, none of the changes in gene expression were
statistically significant. Apparent changes included: Ddit 4, with an apparent up-regulation of 3
A B C
80
fold at 6 h in the first experiment and at 6 h it also showed an apparent up-regulation of 5 fold in
this experiment. Other genes that appeared to be up-regulated, both in the whole ear and peeled
ear experiment, were Mt1a, CEBPd, KIf15, Cyp17a1, Snfilk, etc. However, there were few
changes in gene expression that were consistent for all three experiments, which casts further
doubt on whether these apparent changes were real.
Table 4. Genes with apparent high fold, but statistically non-significant, changes in peeled ear
skin 6 h after NVP treatment. Statistical analysis was performed with one-way ANOVA test.
(Rat 230 2.0 chips; NVP 6 h, n=2; controls, n=2).
Gene Fold Gene FoldDdit4 4.77 Adipoq -2.02Stfa3 3.67 Meox2 -2.02Cebpd 3.62 Sox18 -2.03Dusp1 3.49 Mex3b -2.04RGD135934 3.45 Ccr1 -2.08Tcfcp2l1 3.21 Zfp322a -2.11Mt1a 3.01 Apobec1 -2.12Cebpd 2.94 Nog -2.12Errfi1 2.85 Lipg -2.17Cyp17a1 2.79 Gimap4 -2.19Klf15 2.75 G0s2 -2.21Tsc22d3 2.40 Sele -2.27Klf15 2.23 Serpinb2 -2.28Snf1lk 2.14 Mycn -2.29Mt1a 2.09 LOC500013 -2.35Pfkfb3 2.07 RGD156621 -2.36LOC497995 2.04 Zeb2 -2.47Dusp1 2.00 Hells -2.52
Aplnr -2.54Myh4 -2.55Pcdh18 -2.55Cxcl12 -2.71Evi2a -2.89
The fourth microarray study was performed with peeled ear skin after NVP or DNVP treatment
(Table 5). This study was done with a new chip (Rat ST 1.0 array) because between this and the
81
3 prior studies Affymetrix had changed their rat chip, and the old chip was more expensive and
less complete than the new chip. Again, none of the changes met the criteria for statistical
significance. Apparent changes included up-regulation of keratin genes (Krts) after NVP
treatment, while Nebulin (Neb) and titin (Ttn) appeared to be most down-regulated after DNVP
treatment.
Table 5. Genes with apparent high fold, but statistically non-significant, changes in peeled ear
skin after 6 h NVP (A) or DNVP (B) treatment. The data were analyzed with a one-way
ANOVA test. (Rat ST 1.0 chips, NVP 6 h, n=3; DNVP 6 h, n=4; controls, n=3)
A B
82
6.2 Real-time PCR and protein level study of some genes in the ear
and serum
In further analysis of different groups of microarray data, a comparison of microarray data from
NVP, 12-OH-NVP, and DNVP experiments 6 h after treatment in the ear was performed (Table
6). Some genes, e.g. Mt2A, Mt1a, and Fkbp5, etc were up-regulated in all three drug treatments.
Interestingly, some genes, e.g. Nr4a3, were only up-regulated with NVP and 12-OH-NVP
treatment, but not with DNVP treatment, suggesting that these genes may be associated with
induction of NVP-induced skin rash, but again, none of the changes met the criteria for statistical
significance.
As a check on the microarray data, real-time PCR was performed for Mt2a, Mt1a, and Fkbp5 in
the ear tissue after NVP treatment (Figure 11). The data showed that NVP treatment did not
induce significant changes in Mt2a, Mt1a, or Fkbp5 gene expression in comparison with
controls.
Real-time PCR was also performed for Nr4a3 in rat ear tissue after NVP, 12-OH-NVP, or DNVP
treatment at different time points (Figure 12). The differences between treated and control
animals were inconsistent, again suggesting that apparent changes were not real.
83
Table 6. A comparison of the microarray data from the ear 6 h after NVP (A, taken from Table
2), 12-OH-NVP (B, taken from Table 3), or DNVP treatment (C, taken from Table 5).
Nr4a3: a nuclear receptor subfamily 4, group A, member 3; Mt1a: metallothionein 1a,
Mt 2A: metallothionein 2a; Cebpδ: CCAAT/enhancer binding protein (C/EBP), delta
Fkbp5: FK506 binding protein 5
A B C
84
Figure 11. Real time-PCR study of the expression of Mt1a, Mt2a, Fkbp5, and S100a7a mRNA in
the ear after NVP treatment. The relative concentration is the calibrator-normalized ratio
between the target gene and the reference gene (β2 microglubulin, β2M). Legend: NVP12hr1: rat
No.1 after 12 h NVP treatment and so on; MC12hr1, rat No.1 after 12 h MC treatment; Control1,
rat No.1 non-treated.
85
Figure 12. Real time-PCR study of gene expression (relative concentration) of Nr4a3 in rat ears
6, 24, or 48 h after NVP treatment (A), 6 or 12 h after 12-OH-NVP treatment (B) or 6 h after
NVP treatment (C). Legend: NVP6hr1, rat No.1 after 6 h NVP treatment and so on; Control1, rat
No.1 not treated.
B
A
C
86
Microarray data showed that the S100a7a gene was up-regulated (not significantly) by 2 fold 6
or 12 h after NVP treatment (i.p.) (Table 2 and Figure 13). Real time-PCR data for this gene also
did not show significant changes 12 or 24 h after NVP treatment (i.p.) (Figure 11). Because
S100a7a is an important danger signal, protein expression of this gene in the ear was
investigated. However, western blotting analysis of protein expression of S100a7a in the ear 6,
12, 24, 48, or 72 h after NVP treatment (i.p.) did not find any significant changes, which was
also true for immunohistochemistry analysis of this protein in ear sections 72 h after NVP
treatment (i.p.) (Figure 13).
HMGB1 is also an important danger signal; therefore, even though the fold changes in the
microarray data was only 1.3 fold after NVP treatment (i.p.) (Figure 14), real time-PCR analysis
was also performed, but again the gene was not significantly increased in the ear at 6 h or other
time points, e.g. 12, 24, 48, or 72 h after NVP treatment (i.p.). The level of this protein was also
studied by western blotting (Figure 15); however, western blotting analysis of the HMGB1
protein in the ear lysate from rats 6, 12, 24, 48, or 72 hrs after NVP treatment (i.p.) did not show
significant changes in protein expression. As mentioned in the Introduction, an important
characteristic of HMGB1 is its translocation from nuclei to cytosol when it was acetylated,
which is stimulated by stress such as inflammation. The acetylation of HMGB1 changes its PI,
which can be detected by 2D-electrophoresis and immunoblotting analysis (139). When 2D-
electrophoresis and immunoblotting analysis were applied to determine HMGB1 protein in rat
ear tissue after NVP treatment, no significant PI change of this protein was detected (Figure 15).
In order to further test whether HMGB1 was released into serum after drug treatment, ELISA
was performed on the serum of rats 6 or 12 h after NVP, 12-OH-NVP or DNVP treatment (i.p.).
However, no significant changes were observed (Figure 16).
87
Gene Fold change ( 6 h) Fold change (12 h)
S100a7a 1.7 2.1
Figure 13. The top panel summarizes the microarray data of S100a7a gene expression in the ear
6 or 12 h after NVP treatment. The bottom panel is western blotting analysis and
immunohistochemistry analysis of S100a7a protein in the ear after NVP treatment. For western
blotting, the ear samples were taken from rats 6, 12, 24, 48, or 72 h after NVP treatment or
control animals. Legend: 121, rat No.1 12 hour after NVP treatment; c subscript indicates an
untreated control. mc subscript indicates rats treated with MC. For immunohistochemistry, the
brown color is positive staining for S100a7a, while the blue color is counter staining for nuclei.
The magnifications for these two images were 40x.
88
Gene Fold change (NVP 6 h)
HMGB1 1.3
Figure 14. A summary of microarray analysis (top panel) of HMGB1 gene expression in the ear
6 h after NVP treatment and real time-PCR analysis (bottom panel) of HMGB1 gene expression
in the ear 6, 12, 24, 48, or 72 h after NVP treatment or in control ears. The relative concentration
refers to the fold change of HMGB1 gene expression in comparison with the housekeeping gene,
GAPDH. Legend: NVP6hr1, rat No.1 6 h after NVP treatment and so on; Control1, non-treated
rat No.1.
89
Figure 15. Western blotting and 2D-electrophoretic analysis of HMGB1 protein in the ear after
NVP treatment. The upper panel was for the western blotting analysis of HMGB1 in the samples
taken from homogenized ear tissues of rats 6, 12, 24, 48, or 72 h after NVP treatment or
untreated controls. The lower panel is 2D-electrophoretic and immunoblotting analysis of the
HMGB1 protein in the ear lysate from a rat 24 h after NVP treatment and a control rat. The
circulated bright dots are the HMGB1 protein in the two samples. Legend: 61, rat No.1, 6 h after
NVP treatment and so on; 6mc, a rat after 6h 5% MC treatment; 6c, non-treated rat.
90
Figure 16. ELISA analysis of the HMGB1 protein concentration (ng/mL mean ± s.d, n=4) in rat
serum 6 or 12 h after NVP, 12-OH-NVP or DNVP treatment or from control rats. Legend:
NVP6hr1: rat No.1 6 h after NVP treatment and so on; MC6hr, 6 h after MC treatment.
91
6.3 Changes in gene expression in the liver 6 or 12 h after NVP or
12-OH-NVP treatment
Genes with an statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h after NVP
treatment (gavage) are shown in Table 7A. Interesting genes included those involved in drug
metabolism, e.g. Cyp 2b1, P450 oxidoreductase (Por), NADH dehydrogenase (Ndufaf4); in
immunity, e.g. ζ-associated protein of 70 kDa (Zap70, associated with control of immune
tolerance), FK506 binding protein (Fkbp5, also referred to as immunophilin, involved in
immunoregulation and protein folding), Immunity-related GTPase family M protein (Irgm M,
involved in protein folding), ER degradation enhancer and mannosidase alpha-like 1 (Edem 1,
involved in protein folding and dagradation). The most down-regulated gene was neuronal
regeneration related protein (Nrep) whose function is unknown.
Genes with an statistically significant fold change of ≥ 2 or ≤ -2 in the liver 12 h after NVP
treatment (gavage) (Table 7B) were fewer than after 6 h with more genes involved in
metabolism, e.g. Cyp2b1, indolethylamine N-methyltransferase (Inmt), Cyp3a9, Cyp3a23/3a1
and Por, etc; in energy generation, e.g. ATP-binding cassette, sub-family B (MDR/TAP), and
member 1A (Abcb1a). One remarkable change was Cyp2b1 whose expression increased by 20
fold. Another interesting gene is endothelial cell-specific molecule 1 (Esm1), which increased in
the liver 6 h after 12-OH-NVP treatment but 12 h after NVP treatment (gavage), which suggests
that it was increased by 12-OH-NVP because by 12 h there were significant levels (5 µg/mL) of
12-OH-NVP in NVP-treated animals.
92
Genes with a statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h after 12-OH-NVP
treatment (gavage) are shown in Table 8A. Interesting genes involved in drug metabolism, e.g.,
Cyp2b1, hydroxyprostaglandin dehydrogenase 15 (Hpgd), Por, Cyp4b1, etc.; in immunity, e.g.,
Cd36, CD209b antigen, etc. The fold change of Cyp 2b1 was lower than that after NVP
treatment. The most up-regulated gene was Esm1, and the most down-regulated gene was again
Nrep. Genes with an statistically significant fold change of ≥ 2 or ≤ -2 in the liver 12 h after 12-
OH-NVP treatment (gavage) were fewer than 6 h after 12-OH-NVP treatment with more genes
down-regulated (Table 8B). Some genes had increased expression from 6 to 12 h, while the
expression of other genes decreased. The expression of Cyp2b1 increased from 6 to 12 fold,
while the expression of Esm1 decreased from 6 to 3 fold.
The Cyp2b1 gene was highly induced in the liver by NVP treatment: from 15 fold at 6 h to 20
fold at 12 h; it was also significantly induced by 12-OH-NVP treatment from 6 fold at 6 h to 12
fold at 12 h. Since it is known that NVP is an inducer for Cyp2B6 protein in humans, it is
interesting to compare rat Cyp2B1 amino acid sequence with human Cyp2B6 sequence to
determine their homology. The alignment (Figure 17) showed that the homology is 77.8%.
93
Table 7. Genes with a statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h (A) or 12
h (B) after NVP treatment. The data were analyzed by a two-way ANOVA test. (Rat 230 2.0
chips, NVP 6 h, n=4; controls, n=4; NVP 12 h, n=4; controls, n=4)
Gene Fold Gene FoldCyp2b1 // 14.65 Irgm -2.05Zap70 3.13 Igtp -2.06Por 2.85 Npas2 -2.17Edem1 2.74 Adamts9 -2.18Tsku 2.69 Hao2 -2.18Por 2.57 Ly86 -2.20Fam134b 2.39 Cotl1 -2.26Fkbp5 2.34 P2rx7 -2.29Itpr1 2.33 Ddhd1 -2.37Edem1 2.29 Trim24 -2.55Edem1 2.18 Irs3 -2.81Ndufaf4 2.14 Ppp2r2b -4.12Rhbdd2 2.11 Nrep -6.50
Gene Fold Gene FoldCyp2b1 // 20.43 Pcp4l1 -2.01Inmt 7.83 Cotl1 -2.19Gadd45b 6.23 Cml5 -2.31Rbm3 3.50 Pcp4l1 -2.68Igh-6 /// 3.36Nptx2 3.20Slco1a4 2.73Esm1 2.72Cyp3a9 2.39Abcb1a 2.31Meox2 2.29Insig2 2.26Slco1a4 2.25Cyp3a23/3 2.21Dip2a /// 2.17Insig2 2.15Abcb1a // 2.15Abcb1a 2.13Ces2 2.13Por 2.03
A B
94
Table 8. Genes with a statistically significant fold change of ≥ 2 or ≤ -2 in the liver 6 h (A) or 12
h (B) after 12-OH-NVP treatment. The data were analyzed by a two-way ANOVA test. (Rat 230
2.0 chips, 12-OH-NVP 6 h, n=4; controls, n=4; 12-OH-NVP 12 h, n=4; controls, n=4)
Gene Fold Gene FoldEsm1 5.76 Hao2 -2.04Cyp2b1 // 5.71 Hdc -2.06Krt23 3.16 Prf1 -2.08Cd36 3.11 Adamts9 -2.17Fam134b 2.48 Slc34a2 -2.17Ms4a6b 2.47 Irs3 -2.30Gpr116 2.45 Rprm -2.46Lifr 2.44 Srebf1 -2.52Hpgd 2.33 Npas2 -2.78Por 2.28 Ppp2r2b -3.81Cd209b 2.25 Nrep -8.08Pnpla2 2.23Nampt 2.17Lrg1 2.13Cyp4b1 2.12Mrc1 2.11Cd36 2.10Vcam1 2.08Por 2.01
Gene Fold Gene FoldCyp2b1 // 12.49 Prf1 -2.01Arntl 4.00 Gzma -2.29Pir 3.29 Glul -2.37Esm1 3.22 Cux2 -2.40Hpgd 2.15 Inhbe -2.45
Fmo1 -3.02Nr1d1 -3.30Hsd17b2 -4.04Dbp -4.39
BA
95
1 melsvllfla lltglllllv qrhpnthdrl ppgprplpll gnllqmdrrg llksflrfre
mepsilllla llvgfllllv rghpksrgnf ppgprplpll gnllqldrgg llnsfmqlre
61 kygdvftvhl gprpvvmlcg veairealvd kaeafsgrgk iamvdpffrg ygvifangnr
kygdvftvhl gprpvvmlcg tdtikealvg qaedfsgrgt iaviepifke ygvifanger
121 wkvlrrfsvt tmrdfgmgkr sveeriqeea qclieelrks kgalmdptfl fqsitaniic
wkalrrfsla tmrdfgmgkr sveeriqeea qclveelrks qgapldptfl fqcitaniic
181 sivfgkrfhy qdqeflkmln lfyqtfslis svfgqlfelf sgflkyfpga hrqvyknlqe
sivfgerfdy tdrqflrlle lfyrtfslls sfssqvfeff sgflkyfpga hrqisknlqe
241 inayighsve khretldpsa pkdlidtyll hmekeksnah sefshqnlnl ntlslffagt
ildyighive khratldpsa prdfidtyll rmekeksnhh tvfhhenlmi sllslffagt
301 ettsttlryg fllmlkyphv aervyreieq vigphrppel hdrakmpyte aviyeiqrfs
etssttlryg fllmlkyphv aekvqkeidq vigshrlptl ddrskmpytd aviheiqrfs
361 dllpmgvphi vtqhtsfrgy iipkdtevfl ilstalhdph yfekpdafnp dhfldangal
dlvpigvphr vtkdtmfrgy llpkntevyp ilssalhdpq yfdhpdsfnp ehfldangal
421 kkteafipfs lgkriclgeg iaraelflff ttilqnfsma spvapedidl tpqecgvgki
kkseafmpfs tgkriclgeg iarnelflff ttilqnfsvs shlapkdidl tpkesgigki
481 pptyqirflp r
pptyqicfsa r
Homology between human and rat: (491-109)/491=77.8%
Figure 17. Comparison of the amino acid sequence between human CYP2B6 and
Brown Norway rat CYP2B1 proteins. The bold line is for Brown Norway rats. The highlighted
amino acids are different between the two CYPs.
96
6.4 Changes in gene expression in the skin after NVP or 12-OH-NVP
treatment
Although the first sign of a NVP-induced skin rash in rats is red ears and they do develop a type
of rash on the ear along with the rash on the body, the rash is most pronounced on the back.
Given the minimal findings in samples from the ear, a decision was made to study changes in the
skin from the back. Unlike the data obtained from ear samples, there were 525 genes whose
expression was statistically different between the treatment groups as determined by ANOVA
analysis using Partek genomics suite software. Further comparison between animals from 12-
OH-NVP-treated animals and control showed that 2565 genes were significantly changed, while
no significant changes in gene expression was found in the comparison between NVP-treated
animals and control animals.
Clustering of 525 genes in 12 individual samples showed a distinction between 12-OH-NVP-
and NVP-treated animals in comparison with controls (Figure 18A). In each drug treatment
group, one animal (indicated by an arrow) was significantly different from the other three. After
the outlying individual animals were removed from both 12-OH-NVP and NVP treatment
groups, 2498 statistically significant genes were identified in analysis for the p value of treatment
among the three treatments, e.g. NVP, 12-OH-NVP, and controls (Figure 18B). Further
comparison between 12-OH-NVP-treated animals and control animals showed that 4579 genes
were significantly changed, while 576 genes were significantly changed in comparison between
NVP-treated animals and controls (Figure 18). When the fold change was also considered, the
expression of 442 genes were significantly changed ≥ 2 or ≤ -2 fold in comparison between 12-
97
OH-NVP-treated animals and controls, while the expression of only 43 genes was changed in
comparison between NVP-treated animals and controls.
Examples of significant changes in gene expression in the skin after 12-OH-NVP treatment are
shown in Table 9. The gene with the greatest change (18 fold) was Trim63, an ubiquitin ligase,
which is involved in the protein folding. Another gene that may be involved in protein folding
was FK506 binding protein 5 (Fkbp5). Among other genes, some were associated with immune
response, e.g. IL-22ra2 (soluble IL-22 receptor and an antagonist for IL-22) and S100a7a (a
documented danger signal that is involved in the pathogenesis of psoriasis). Some of the genes
are expressed in the mitochondria, e.g. pyruvate dehydrogenase kinase, isozyme 4 (Pdk4) and 3-
hydroxy-3-methylglutaryl-coenzyme A synthase 2 (Hmgcs2) and uncoupling protein 3 (Ucp3),
etc.; some are associated with apoptosis, e.g. death associated protein kinase 1 (Dapk1) and
Kruppel-like factor 15 (Klf 15), etc.; and some were associated with cell stress, e.g. lipin1
(Lpin1), DnaJ (Hsp40) homolog, subfamily B, member 5 (Dnajb5), etc.
There were far fewer significant changes in gene expression in the skin of NVP-treated animals
as shown in Table 10. The gene with the greatest change (4 fold) was Hmgcs2, which is involved
with mitochondrial function. No immune response or stress-related genes were found in this list.
Fold changes of gene expression in the skin 6 h after both 12-OH-NVP and NVP treatment are
listed in Figure 19A. Fold changes of Trim 63 and S100a7a after NVP treatment were 5.3 and
4.0, respectively; however, neither was statistically significant. When changes in gene expression
from the skin of 12-OH-NVP-treated animals were analyzed by Ingenuity pathway software, the
top network was “Cell-To-Cell Signaling and Interaction, Tissue Development and
Hematological System Development and Function” (Figure 19B). However, it is difficult to draw
98
any conclusions from this analysis because the genes with the greatest change in expression were
not included in this network.
Figure 18. A: Clustering of 525 genes from the one-way ANOVA analysis for statistically
significant genes among three drug (NVP, 12-OH-NVP, and MC control) treatment groups (p
value of treatment with FDR < 0.05) in 12 skin samples; B: A summary of a further one-way
ANOVA analysis for the p value among the three drug treatment groups in 10 skin samples (one
sample taken out from each NVP and 12-OH-NVP treatment group). The sample that was
removed is indicated by an arrow at the top of the heat map.
A B
99
Table 9. Examples of genes with a significant change in gene expression in the skin 6 h after 12-
OH-NVP treatment. The data were analyzed by a one-way ANOVA. (Rat 230 2.0 chips; 12-OH-
NVP, n=4; controls, n=4)
100
Table 10. Examples of genes with a significant change in gene expression in the skin 6 h after
NVP treatment. The data were analyzed by a one-way ANOVA test. (Rat 230 2.0 chips, NVP,
n=4; controls, n=4)
Gene Fold Gene FoldHmgcs2 3.90 Clec4a1 -2.17Fibin 2.80 Ms4a7 -2.19Fam107a 2.56 Clec4a3 -2.26Tox3 2.23 Slamf9 -2.27Mgp 2.06 Lilrb3l -2.39Tsc22d3 2.04 Lilrb3l / -2.39Pik3ip1 2.02 Cotl1 -2.41
Epb4.1l3 -2.49Vsnl1 -2.89Armcx2 -2.91Lipg -3.54
101
Figure 19. A. Fold changes in gene expression in the skin 6 h after 12-OH-NVP or NVP
treatment. B. The pathway analysis of genes with changes in expression after 12-OH-NVP
treatment using Ingenuity software.
102
6.5 Blood levels of IL-22ra2 and S100a7a protein in the skin
The serum levels of IL-22ra2 during 8 days of NVP or 12-OH-NVP treatment in food was
analyzed with ELISA and shown in Figure 20A. When rats were treated with NVP, IL-22ra2
levels in the serum fluctuated between 0.6 ng/mL and 1.7 ng/mL. When rats were treated with
12-OH-NVP in food, IL-22ra2 levels in the serum fluctuated between 0.8 ng/mL and 1.2 ng/mL.
No significant changes were found after drug treatment, and no significant difference was found
between the two drug treatments.
The serum level of NVP, 12-OH-NVP metabolite after NVP treatment, and 12-OH-NVP after
treatment with 12-OH-NVP from the same experiment is shown in Figure 20B. At day 4, the
concentration of both NVP (NVP treatment) and 12-OH-NVP (12-OH-NVP treatment) were 25
µg/mL. While after day 4, the NVP level was still increasing while the 12-OH-NVP level had
started to decrease, and at day 8, NVP and 12-OH-NVP levels were very different. The NVP
level was 45 µg/mL, while the 12-OH-NVP level was only 10 µg/mL. From day 2 to day 8, the
12-OH-NVP metabolite blood levels in NVP-treated rats stayed between 10 and 13 µg/mL,
Western blotting analysis of S100a7a expression in the skin after same NVP or 12-OH-NVP
treatment in food for 8 days did not show any significant changes (Figure 20).
103
Figure 20. A. The top panel is the serum level of IL-22ra2 in rats after NVP or 12-OH-NVP
treatment. B. The bottom panel is the serum level of NVP, 12-OH-NVP metabolite and 12-OH-
NVP in rats after NVP (n=2) or 12-OH-NVP (n=4) treatment in food in the same experiment.
A
B
104
Figure 21. Western blotting analysis of S100a7a expression in rat skin after NVP (n=4) or 12-
OH-NVP (n=4) treatment in food for 8 days. The samples (labeled 1, 2, 3, 4) were rat skin
lysates from 4 rats after NVP or 12-OH-NVP treatment.
105
7 Discussion
The aim of these studies was to determine the effect of NVP on expression of mRNAs in the skin
and liver in the NVP-induced skin rash rat model to determine if there were changes consistent
with the Danger Hypothesis. In addition, the association between the metabolism of NVP in both
the liver and the skin, and danger signal induction in these two organs was also studied.
Our lab had previously determined that the NVP metabolite, 12-OH-NVP, is responsible for the
skin rash in the NVP animal model, and substitution of deuterium for hydrogen on the methyl
group to form DNVP, which inhibits the formation of 12-OH-NVP, decreased the incidence and
severity of the NVP-induced skin rash. In this study, 12-OH-NVP was used as a positive control,
and DNVP was used as a negative control to determine if there were danger signals in the animal
model. However, the microarray screening study in rat ear, including whole ear and peeled ear
skin after NVP, 12-OH-NVP, or DNVP treatment did not detect any significant changes in gene
expression in the ear. The reasons may be due to the small (n=2) number of animals at each time
point (6 or 12 h) of the two drug treatments. It could be that there are no changes in the ear or
changes were different than those from back because we also speculate that the ear may be
different from skin of the back in terms of histological structure and drug metabolism, such as
sulfotransferases. The sulfotransferases are reported to be located in the epidermis (252) and hair
follicles (197) in the skin, and our lab has found covalent binding in the epidermis but not in
dermis of NVP-treated rat skin (unpublished data). In the ear, sulfotransferases may be scarce,
and further investigation is needed.
Some genes in the ear (Table 2-5) appeared to have relatively large fold changes in expression.
Although the changes were not statistically significant, they could generate hypotheses to test;
106
however, they also were not consistent between experiments so it is unlikely that they were real.
Although we did not see significant changes for S100a7a and HMGB1 gene expression, the two
documented danger signals in the ear protein levels are more important. Protein level studies of
S100a7a and HMGB1 in the ear after NVP treatment with both western blotting and
immunohistochemistry technologies found no significant changes (Figures 13 and 16). In
particular, the activity of HMGB1 is based on posttranslational modifications that lead to exit
from the nucleus (140). However, even 2-dimensional electrophoresis looking for changes in
HMGB1 acetylation failed to detect clear changes (Figure 15).
Although there was clearly an infiltration of lymphocytes in the ears of NVP-treated animals
with a skin rash (235, 243), we did not see any rash on the ears, even during secondary treatment,
and that may explain why we did not detect any significant changes in gene expression in the ear
after NVP or 12-OH-NVP treatment.
As we did not successfully find danger signals in the ear, the exploration for danger signals was
switched to the liver and other skin. The major metabolic site of reactive metabolite formation is
the liver. The list of genes with statistically significant (FDR<0.05) changes in expression in the
liver produced by NVP or 12-OH-NVP treatment were different (Tables 7 and 8). This result was
consistent with what we know about NVP and 12-OH-NVP bioactivation (Figure 9): NVP can be
directly oxidized to the reactive quinone methide and this appears to be the major source of
covalent binding in the liver, while 12-OH-NVP cannot be oxidized to the quinone methide, but
it can be metabolized to the less reactive sulfate. There may also be other minor pathways
leading to a reactive metabolite. In this study, NVP induced more changes in gene expression
than 12-OH-NVP, especially at 12 h (Tables 7 and 8). One of the liver metabolic enzymes, i.e.
CYP 2B1, was greatly up-regulated by both NVP and 12-OH-NVP treatment at 6 and 12 h. This
107
is consistent with studies in which NVP has been found to be an inducer of CYP2B6 in humans
(253) and CYP2B1 in rats (10). An amino acid sequence comparison (Figure 17) showed high
homology (77.8%) between human CYP2B6 and rat CYP2B1, suggesting that they are
homologous. In humans, NVP was found to induce CYP3A4 and CYP2B6 through activation of
constitutive human androstane receptor (hCAR) but not human pregnane X receptor (hPXR)
(254). However, no change in CAR gene expression was detected in the liver of NVP- or 12-OH-
NVP-treated rats. The time of their activation may be much earlier than 6 h and this could be
tested.
Similar liver enzymes are induced in both humans and rats by NVP treatment, but most humans
and rats do not develop liver toxicity. This may be associated with the ZAP70 (ζ-associated
protein of 70 kDa) tyrosine kinase gene, which was found to be induced in the liver of rats by
NVP treatment. ZAP-70, engaged with T-cell receptors, plays a critical role in activating many
downstream signal transduction pathways in T cells and can result in both positive and negative
selection (255). Data from a recent study has shown that a defect in ZAP-70 resulted in
immunodeficiency, ultimately resulting in autoimmunity, which suggests that ZAP-70 is
associated with control of immune tolerance (256-257). This may explain why these animals do
not develop hepatotoxicity (256-257).
Other potential danger signal genes (Table 7) in the liver after NVP treatment that were not
induced by 12-OH-NVP treatment included: Por, TSKU, Gadd45b, Edem1, and Fkbp5. Por is
P450 cytochrome reductase, which is required to reduce P450 enzymes in the liver. Tsku (Eiih)
is also referred to as early insulin-induced hepatic gene. Although its function is unclear, it is
induced in the liver by several aromatic amine drugs, specifically aminoglutethimide,
sulfamethoxazole and dapsone (unpublished data from our lab). Gadd45b, growth arrest and
108
DNA-damage-inducible β gene, has been found as a novel mediator of apoptosis in
cardiomyocytes in response to ischemia/hypoxia (258-261). Edem1, ER degradation-enhancing
alpha-mannosidase-like 1, is associated endoplasmic reticulum stress and is a receptor of
terminal misfolded proteins from the ER (262-263). Fkbp5 forms a complex with Hsp90 (264)
and is involved in immunosuppression and protein folding (265-266).
The two most down-regulated genes in the liver that resulted from NVP treatment (Table 7) were
Ppp2r2b and Nrep. Ppp2r2b, a member of protein phosphatase 2 family, is down-regulated in
Alzheimer’s disease (267-268). Nrep, a neuronal regeneration-related protein (also referred to as
P311), was found to be down-regulated by TGF-ß 1 and 2 in vitro (269) and is involved in facial
nerve regeneration in vivo (270). However, it is unclear how these two genes might be related to
the NVP-induced skin rash.
The list of genes with a significant (FDR<0.05) change in mRNA expression in the liver 12 h
after NVP and 12-OH-NVP treatments (Tables 7 and 8) was shorter than the list 6 h after
treatment, indicating that the earlier time point is probably optimal for detecting danger signals,
although several time points may detect more danger signals because some gene changes may be
very transient.
In contrast to NVP whose reactive metabolite, a quinone methide, can covalently bind to proteins
in the liver to induce gene changes in protein folding, 12-OH-NVP treatment did not induce
protein folding-related gene changes in the liver (Table 8). 12-OH-NVP cannot directly form the
quinone methide, because they are at same oxidation state. Although 12-OH-NVP can
theoretically be oxidized to a relatively reactive aldehyde, it appears that the aldehyde is directly
oxidized to the carboxylic acid without leaving the P450 active site because it was not detected
in incubations of 12-OH-NVP with hepatic enzymes even though the carboxylic acid was readily
109
detected (unpublished data). The carboxylic acid is not reactive, but it can from an acyl
glucuronide that has the potential to bind to proteins in the liver. However, in previous studies
aminobenzotriazole markedly decreased the concentration of the carboxylic acid, but increased
the incidence of rash.
The difference in changes in gene expression induced in the skin by NVP or 12-OH-NVP
treatment greatly supports the hypothesis that the 12-OH-NVP pathway is responsible for the
skin rash. The number of significantly regulated genes in the skin by 12-OH-NVP was much
greater than that by NVP treatment (Appendix 1 and 2, Figure 18), which is consistent with the
hypothesis that reactive metabolites in the skin derived from 12-OH-NVP is responsible for the
rash. The most likely reactive metabolite is the sulfate (Figure 10), and preliminary inhibition
studies indicate that sulfate formed in the skin and not the liver is responsible for the rash.
Ingenuity software was used to analyze possible networks or pathways that are associated with
the genes for which there were significant changes in expression in rat skin after 12-OH-NVP
treatment. However, no clear pathways or networks were identified that would likely lead to an
immune response (Figure 19). One possible reason is that the number of skin genes in the data-
base of Ingenuity software is not big enough to build reliable pathways or networks.
NVP-induced skin rash is an immune-mediated reaction; therefore, IL-22ra2 and S100a7a could
be relevant genes. IL-22ra2 is a soluble receptor for IL-22, an inflammatory cytokine in the IL-
10 family, and also polymorphism of this gene has been associated with the risk of multiple
sclerosis, which is an autoimmune disease (271-272). S100a7a binds to the receptor for advanced
glycation end products (RAGE) and has been found to be up-regulated in skin lesions of patients
with psoriasis (273-274). When 12-OH-NVP is metabolized in the skin and covalently binds in
the skin, IL-22ra2 and S100a7a could be released and circulate in the blood or interact with
110
RAGE on APC cells in the skin, respectively, to initiate an immune response. Other protein
folding-, stress-, and apoptosis-related genes such as Trim63, a ubiqutin ligase associated with
protein turn-over in muscle (275); Fkbp5, an immunophilin involved in immunosuppression and
protein folding (265); Pdk4, pyruvate dehydrogenase kinase 4, induced by hepatotoxins in the
liver and regulated by Cebpb (276-277); Dapk1, death associated protein kinase 1, involved in
apoptosis and autophagy (278-280); Lpin1, phosphatidic acid phosphatase for production of 1,2-
diacylglycerol, induced by hypoxia and stress (281); Ucp3, mitochondrial uncoupling protein 3,
involved in oxidative metabolism of fatty acids and induced by hypoxia-induced oxidative stress
(282-283); Dnajb5, DnaJ (Hsp40) homolog, subfamily B, member 5, a crucial chaperone for
Hsp70 (284-285); Klf15, Kruppel-like factor 15, involved in oxidative stress (286); Nox4,
NADPH oxidase 4, involved in ER stress-induced caspase-3 activation in vitro (287); Cebpd,
CCAAT/enhancer binding protein delta, involved in DNA damage (288) and induced to assist
protein folding and correcting stress or to signal stress to induce apoptosis. All these genes are
significantly up-regulated by 12-OH-NVP treatment, but their expression was not significantly
changed by NVP treatment. They can be considered to be potential danger signals because they
are either associated with protein folding, stress, or apoptosis.
As mentioned above, because IL-22ra2 is a soluble receptor for IL-22 (272), and the IL-22ra2
gene was significantly up-regulated by 12-OH-NVP treatment, we wanted to determine whether
IL-22ra2 protein was released into serum during NVP or 12-OH-NVP treatment. However, no
significant change was found during 8 days NVP or 12-OH-NVP in food treatment (Figure 20),
indicating that IL-22ra2 may serve as a danger signal in the skin instead of being released into
serum.
111
S100a7a is a danger signal and its mRNA expression was significantly up-regulated in the skin
after 12-OH-NVP treatment. S100 protein family members, e.g. S100a7, S100a8, and S100a9,
etc., have been found to be highly up-regulated in two in vivo microarray analyses of gene
changes in the skin from The National Center for Biotechnology Information (NCBI) Gene
Expression Omnibus (GEO) data base: ‘’Gene expression data of skin from psoriatic patients and
normal controls”(289) and “Gene Expression Time Course in the Human Skin during Elicitation
of Allergic Contact Dermatitis”(290). In order to determine protein levels of S100a7a, rat skin
was taken for western blotting analysis 8 days after NVP or 12-OH-NVP treatment; however, no
significant changes of S100a7a protein expression was found in the skin between these two drugs
(Figure 21). Eight days may not be the best time point for protein level analysis of S100a7a
because the mRNA was up-regulated in the skin 6 h after 12-OH-NVP treatment. It is also
possible that the change was localized to one part of the skin, such as the epidermis, which is the
location of covalent binding but very thin compared to the dermis. Therefore, western blotting of
the whole skin may not be able to detect changes localized in the epidermis. Therefore,
separating the dermis and epidermis and testing them separately might increase the sensitivity.
Pycr1, pyrroline-5-carboxylate reductase 1, was the most down regulated gene in rat skin by 12-
OH-NVP treatment (Table 9). Pycr1 catalyzes the conversion of pyrroline-5-carboxylate to
proline. A recent study found that a mutation in Pycr1 gene caused an autosomal recessive cutis
laxa in a family, which was characterized by wrinkled, redundant, inelastic, and sagging skin due
to defective synthesis of elastic fibers and other proteins of the extracellular matrix (291).
Although it seems that down regulation of Pycr1 in the skin from 12-OH-NVP treatment may be
associated with the skin rash, it is not clear how this gene would be related to the initiation of an
immune response.
112
In contrast, the number of significant genes induced by NVP in the skin is much shorter than that
from 12-OH-NVP treatment (Table 10), and no significant genes were found to be associated
with protein-folding, stress, or immune responses, further suggesting the role of 12-OH-NVP in
NVP-induced skin rash. A recent study by Park’s group investigated bioactivation of NVP in
humans and different rat strains (6). The study confirmed formation of 12-sulfoxy-NVP in rats:
the metabolite was detected in rat urine and bile samples, but the metabolite was not found in
human urine samples. However, the fact that 12-sulfoxy-NVP was not detected in the human
samples does not mean that this metabolite is not produced at all. 12-sulfoxy-NVP was detected
in bile of the treated rats, both by our group (5) and Park’s group (6). Because 12-OH-NVP is
readily oxidized to 4-carboxyl-NVP in the liver and excreted in urine via glucuronidation,
sulfation of 12-OH-NVP is more important in the skin. Our lab has found compelling evidence
that sulfate is formed in the epidermis and covalently binds in the skin; therefore, finding the
sulfate in urine may be irrelevant (unpublished data).
In summary, these studies found changes in mRNA levels in the skin of rats treated with 12-OH-
NVP that likely represent danger signals. The difference between the number and type of
changes induced by 12-OH-NVP vs NVP were striking and are consistent with the observation
that 12-hydroxylation is required to induce the skin rash. It also provides clues to which changes
may be important for induction of an immune response. Although NVP is obviously converted to
12-OH-NVP, presumably at the 6 h time point the amount formed is insufficient to induce the
changes observed with 12-OH-NVP treatment. Another striking observation was the difference
in changes observed in skin from the ear when compared with skin from the back even though
there were histological inflammatory changes in the ear as well as in the skin from the back. It is
also striking that, although the amount of covalent binding in the liver is much greater than in the
skin, the number of changes in gene expression were much smaller.
113
Finding a large number of changes in gene expression in the skin that are induced by 12-OH-
NVP is only the first step. The next challenge is to test which, if any, of these changes lead to an
immune-mediated skin rash. Recent studies in our lab have demonstrated that topical 1-phenyl-1-
hexanol, a sulfotransferase inhibitor, prevented the rash, but only in the area where it was
applied. Future studies can focus on looking for changes in gene expression that are blocked by
topical 1-phenyl-1-hexanol. Also the upstream and down-stream signals for those significant
genes in the skin after 12-OH-NVP treatment can be investigated. For example, if we can knock
out the gene of the receptor of S100a7a, e.g. RAGE, we can determine whether this changes the
incidence of skin rash. Since phorbol esters was found to induce S100a7a gene in keratinocytes
in vitro (273-274), we can use this reagent to induce S100a7a in rats to see whether we can
induce skin rash at a lower dose of NVP or 12-OH-NVP. However, phorbol esters are known to
have many effects so the results of such studies would be difficult to interpret. The danger
signals found in this study may act as a biomarker for other drugs that could cause a serious skin
rash or other type of IDR, although it is likely that the danger signals will be different for
different drugs and in different organs.
114
References 1. Shenton JM, Chen J, Uetrecht JP. 2004. Chem Biol Interact 150: 53-70
2. Product Monograph on use of VIRAMUNE® in the treatment of adults and children with
HIV infection, Boehringer Ingelheim Inc. 2008.
3. Khavari PA. 2006. Nat Rev Cancer 6: 270-80
4. Glatt H. 1997. FASEB J 11: 314-21
5. Chen J, Mannargudi BM, Xu L, Uetrecht J. 2008. Chem Res Toxicol 21: 1862-70
6. Srivastava A, Lian LY, Maggs JL, Chaponda M, Pirmohamed M, et al. 2010. Drug Metab Dispos 38: 122-32
7. Antiretroviral therapy for HIV infection in adults and adolescents, Recommendations for a public health approach, 2010 revision, World Health Organization. 2010.
8. Bacolla A, Shih CK, Rose JM, Piras G, Warren TC, et al. 1993. J Biol Chem 268: 16571-7
9. Smerdon SJ, Jager J, Wang J, Kohlstaedt LA, Chirino AJ, et al. 1994. Proc Natl Acad Sci U S A 91: 3911-5
10. Cheeseman SH, Hattox SE, McLaughlin MM, Koup RA, Andrews C, et al. 1993. Antimicrob Agents Chemother 37: 178-82
11. Murphy R, Montaner J. 1996. Exp. Opin. Invest. Drugs 5(9): 16
12. Guay LA, Musoke P, Fleming T, Bagenda D, Allen M, et al. 1999. Lancet 354: 795-802
13. Jackson JB, Musoke P, Fleming T, Guay LA, Bagenda D, et al. 2003. Lancet 362: 859-68
14. Riska P, Lamson M, MacGregor T, Sabo J, Hattox S, et al. 1999. Drug Metab Dispos 27: 895-901
15. Havlir D, Cheeseman SH, McLaughlin M, Murphy R, Erice A, et al. 1995. J Infect Dis 171: 537-45
16. Lamson MJ, Sabo JP, MacGregor TR, Pav JW, Rowland L, et al. 1999. Biopharm Drug Dispos 20: 285-91
17. Lamson M, Cort S, Sabo J, Keirns J. 1995. Pharmacol Res 12: 1
18. Riska PS, Joseph DP, Dinallo RM, Davidson WC, Keirns JJ, Hattox SE. 1999. Drug metabolism and disposition: the biological fate of chemicals 27: 1434-47
115
19. Erickson DA, Mather G, Trager WF, Levy RH, Keirns JJ. 1999. Drug metabolism and disposition: the biological fate of chemicals 27: 1488-95
20. Pollard RB, Robinson P, Dransfield K. 1998. Clin Ther 20: 1071-92
21. Uetrecht JP. 2000. Curr Drug Metab 1: 133-41
22. Taiwo BO. 2006. Int J STD AIDS 17: 364-9; quiz 70
23. van Leth F, Phanuphak P, Ruxrungtham K, Baraldi E, Miller S, et al. 2004. Lancet 363: 1253-63
24. Manosuthi W, Sungkanuparph S, Tansuphaswadikul S, Inthong Y, Prasithsirikul W, et al. 2007. Int J STD AIDS 18: 782-6
25. Fagot JP, Mockenhaupt M, Bouwes-Bavinck JN, Naldi L, Viboud C, Roujeau JC. 2001. AIDS 15: 1843-8
26. Bersoff-Matcha SJ, Miller WC, Aberg JA, van Der Horst C, Hamrick Jr HJ, et al. 2001. Clin Infect Dis 32: 124-9
27. Wong KH, Chan KC, Lee SS. 2001. Clin Infect Dis 33: 2096-8
28. Gangar M, Arias G, O'Brien JG, Kemper CA. 2000. Ann Pharmacother 34: 839-42
29. Johnson S, Baraboutis JG. 2000. JAMA 284: 2722-3
30. Uetrecht J. 2007. Annu Rev Pharmacol Toxicol 47: 513-39
31. Pirmohamed M, James S, Meakin S, Green C, Scott AK, et al. 2004. BMJ 329: 15-9
32. Uetrecht J. 2008. Chem Res Toxicol 21: 84-92
33. Lee WM. 2003. N Engl J Med 349: 474-85
34. Lasser KE, Allen PD, Woolhandler SJ, Himmelstein DU, Wolfe SM, Bor DH. 2002. JAMA 287: 2215-20
35. Lammert C, Einarsson S, Saha C, Niklasson A, Bjornsson E, Chalasani N. 2008. Hepatology 47: 2003-9
36. Zhang X, Liu F, Chen X, Zhu X, Uetrecht J. 2011. Drug Metab Pharmacokinet 26: 47-59
37. Kanitakis J. 2002. Eur J Dermatol 12: 390-9; quiz 400-1
38. Stenn KS, Paus R. 2001. Physiol Rev 81: 449-94
39. Menon GK. 2002. Adv Drug Deliv Rev 54 Suppl 1: S3-17
40. Zenz R, Eferl R, Kenner L, Florin L, Hummerich L, et al. 2005. Nature 437: 369-75
116
41. Berking C, Takemoto R, Binder RL, Hartman SM, Ruiter DJ, et al. 2002. Carcinogenesis 23: 181-7
42. Nigen S, Knowles SR, Shear NH. 2003. J Drugs Dermatol 2: 278-99
43. Torres MJ, Mayorga C, Blanca M. 2009. J Investig Allergol Clin Immunol 19: 80-90
44. Fernandez TD, Canto G, Blanca M. 2009. Curr Opin Infect Dis 22: 272-8
45. Pichler WJ. 2003. Ann Intern Med 139: 683-93
46. Posadas SJ, Padial A, Torres MJ, Mayorga C, Leyva L, et al. 2002. J Allergy Clin Immunol 109: 155-61
47. Yawalkar N, Egli F, Hari Y, Nievergelt H, Braathen LR, Pichler WJ. 2000. Clin Exp Allergy 30: 847-55
48. Chen X, Tharmanathan T, Mannargudi B, Gou H, Uetrecht JP. 2009. J Pharmacol Exp Ther 331: 836-41
49. Edwards RG, Youlten LJ, Dewdney JM. 1986. Indian J Pediatr 53: 37-44
50. Levine BB. 1964. Immunology 7: 527-41
51. Gueant-Rodriguez RM, Romano A, Beri-Dexheimer M, Viola M, Gaeta F, Gueant JL. 2006. Pharmacogenet Genomics 16: 713-9
52. Mathelier-Fusade P. 2006. Clin Rev Allergy Immunol 30: 19-23
53. Uetrecht J. 2009. Chem Res Toxicol 22: 24-34
54. Picard D, Janela B, Descamps V, D'Incan M, Courville P, et al. 2010. Sci Transl Med 2: 46ra62
55. Valeyrie-Allanore L, Sassolas B, Roujeau JC. 2007. Drug Saf 30: 1011-30
56. Roujeau JC. 2005. Toxicology 209: 123-9
57. Sehgal VN, Srivastava G. 2006. Int J Dermatol 45: 897-908
58. Mockenhaupt M. 2009. J Dtsch Dermatol Ges 7: 142-60; quiz 61-2
59. Abe R. 2008. J Dermatol Sci 52: 151-9
60. Nassif A, Moslehi H, Le Gouvello S, Bagot M, Lyonnet L, et al. 2004. J Invest Dermatol 123: 850-5
61. Chung WH, Hung SI, Chen YT. 2010. Expert Opin Drug Saf 9: 15-21
117
62. Hung SI, Chung WH, Liou LB, Chu CC, Lin M, et al. 2005. Proc Natl Acad Sci U S A 102: 4134-9
63. Temple RJ, Himmel MH. 2002. JAMA 287: 2273-5
64. Ostapowicz G, Fontana RJ, Schiodt FV, Larson A, Davern TJ, et al. 2002. Ann Intern Med 137: 947-54
65. Lawrenson RA, Seaman HE, Sundstrom A, Williams TJ, Farmer RD. 2000. Drug Saf 23: 333-49
66. Zimmerman H. 1999. Hepatotoxicity: the adverse effects of drugs and other chemicals on the liver. Philadelphia: Lippincott Williams and Wilkins
67. Obermayer-Straub P, Strassburg CP, Manns MP. 2000. Can J Gastroenterol 14: 429-39
68. Kleiner DE. 2009. Semin Liver Dis 29: 364-72
69. Uetrecht J. 2009. Semin Liver Dis 29: 383-92
70. Chien RN, Sheen IS, Liaw YF. 2003. Int J Clin Pract 57: 829-30
71. Maddrey WC, Boitnott JK. 1973. Ann Intern Med 79: 1-12
72. Aster RH. 2010. Handb Exp Pharmacol: 57-76
73. Salama A. 2009. Expert Opin Drug Saf 8: 73-9
74. Garratty G. 1993. Transfus Med Rev 7: 255-67
75. Garratty G. 2005. Semin Hematol 42: 119-21
76. Garratty G, Arndt PA. 2007. Immunohematology 23: 105-19
77. Kenney B, Stack G. 2009. Arch Pathol Lab Med 133: 309-14
78. Aster RH, Bougie DW. 2007. N Engl J Med 357: 580-7
79. Warkentin TE. 2003. Br J Haematol 121: 535-55
80. Warkentin TE, Kelton JG. 2001. N Engl J Med 344: 1286-92
81. Aster RH. 2000. Semin Hematol 37: 229-38
82. Andres E, Zimmer J, Affenberger S, Federici L, Alt M, Maloisel F. 2006. Eur J Intern Med 17: 529-35
83. Tesfa D, Keisu M, Palmblad J. 2009. Am J Hematol 84: 428-34
84. Madison. TLSaFW. 1934. Journal of Allergy 6: 8
118
85. Moeschlin S, Wagner K. 1952. Acta Haematol 8: 29-41
86. Fibbe WE, Claas FH, Van der Star-Dijkstra W, Schaafsma MR, Meyboom RH, Falkenburg JH. 1986. Br J Haematol 64: 363-73
87. Akamizu T, Ozaki S, Hiratani H, Uesugi H, Sobajima J, et al. 2002. Clin Exp Immunol 127: 92-8
88. Stroncek DF, Herr GP, Maguire RB, Eiber G, Clement LT. 1994. Transfusion 34: 980-5
89. Safferman AZ, Lieberman JA, Alvir JM, Howard A. 1992. Lancet 339: 1296-7
90. Uetrecht JP. 1992. Drug Saf 7 Suppl 1: 51-6
91. Gardner I, Leeder JS, Chin T, Zahid N, Uetrecht JP. 1998. Mol Pharmacol 53: 999-1008
92. Yunis JJ, Corzo D, Salazar M, Lieberman JA, Howard A, Yunis EJ. 1995. Blood 86: 1177-83
93. Iverson S, Kautiainen A, Ip J, Uetrecht JP. 2010. Chem Res Toxicol
94. Young NS. 1997. Ann Intern Med 126: 166-8
95. Young NS, Scheinberg P, Calado RT. 2008. Curr Opin Hematol 15: 162-8
96. Young NS. 2002. Ann Intern Med 136: 534-46
97. Hoffman R, Zanjani ED, Lutton JD, Zalusky R, Wasserman LR. 1977. N Engl J Med 296: 10-3
98. Nissen C, Cornu P, Gratwohl A, Speck B. 1980. Br J Haematol 45: 233-43
99. Uetrecht J. 1990. Crit Rev Toxicol 20: 213-35
100. Borchers AT, Keen CL, Gershwin ME. 2007. Ann N Y Acad Sci 1108: 166-82
101. Pichler WJ. 2003. Curr Opin Allergy Clin Immunol 3: 249-53
102. Fritzler MJ. 1994. Lupus 3: 455-9
103. Wiik A. 2008. Curr Opin Rheumatol 20: 35-9
104. Vasoo S. 2006. Lupus 15: 757-61
105. Crowson AN, Brown TJ, Magro CM. 2003. Am J Clin Dermatol 4: 407-28
106. Ramos-Casals M, Roberto Perez A, Diaz-Lagares C, Cuadrado MJ, Khamashta MA. 2010. Autoimmun Rev 9: 188-93
119
107. Woosley RL, Drayer DE, Reidenberg MM, Nies AS, Carr K, Oates JA. 1978. N Engl J Med 298: 1157-9
108. Yung RL, Johnson KJ, Richardson BC. 1995. Lab Invest 73: 746-59
109. Sawalha AH, Richardson BC. 2007. Systemic lupus erythematosus: a companion to Rheumatology Philadelphia: MOSBY ELSEVIER. 10 pp.
110. Uetrecht J. 2005. Autoimmun Rev 4: 309-14
111. Li J, Mannargudi B, Uetrecht JP. 2009. Chem Res Toxicol 22: 1277-84
112. Li J, Uetrecht JP. 2009. Chem Res Toxicol 22: 1526-33
113. Dedeoglu F. 2009. Curr Opin Rheumatol 21: 547-51
114. Walgren JL, Mitchell MD, Thompson DC. 2005. Crit Rev Toxicol 35: 325-61
115. Hyson C, Sadler M. 1997. Can J Neurol Sci 24: 245-9
116. Shear NH, Spielberg SP. 1988. J Clin Invest 82: 1826-32
117. Alvir JM, Lieberman JA, Safferman AZ, Schwimmer JL, Schaaf JA. 1993. N Engl J Med 329: 162-7
118. Walton B, Simpson BR, Strunin L, Doniach D, Perrin J, Appleyard AJ. 1976. Br Med J 1: 1171-6
119. Mallal S, Nolan D, Witt C, Masel G, Martin AM, et al. 2002. Lancet 359: 727-32
120. Martin AM, Nolan D, Gaudieri S, Almeida CA, Nolan R, et al. 2004. Proc Natl Acad Sci U S A 101: 4180-5
121. Landsteiner K, Jacobs J. 1935. J Exp Med 61: 643-56
122. Landsteiner K, Pauling L. 1945. The specificity of serological reactions. Rev. ed. With a chapter on molecular structure and intermolecular forces by Linus Pauling. Cambridge: Harvard University Press. 310p. pp.
123. Janeway C. 2005. Immunobiology : the immune system in health and disease. New York: Garland Science. xxiii, 823 p. pp.
124. Parker CW, Deweck AL, Kern M, Eisen HN. 1962. J Exp Med 115: 803-19
125. Brodie BB, Reid WD, Cho AK, Sipes G, Krishna G, Gillette JR. 1971. Proc Natl Acad Sci U S A 68: 160-4
126. Mitchell JR, Jollow DJ, Gillette JR, Brodie BB. 1973. Drug Metab Dispos 1: 418-23
120
127. Vergani D, Mieli-Vergani G, Alberti A, Neuberger J, Eddleston AL, et al. 1980. N Engl J Med 303: 66-71
128. Uetrecht JP. 1999. Chem Res Toxicol 12: 387-95
129. Medzhitov R, Janeway CA, Jr. 1997. Curr Opin Immunol 9: 4-9
130. Matzinger P. 1994. Annu Rev Immunol 12: 991-1045
131. Seguin B, Uetrecht J. 2003. Curr Opin Allergy Clin Immunol 3: 235-42
132. Levy M. 1997. Drug Saf 16: 1-8
133. Pullen H, Wright N, Murdoch JM. 1967. Lancet 2: 1176-8
134. Fischl MA, Dickinson GM, La Voie L. 1988. JAMA 259: 1185-9
135. Ellrodt AG, Murata GH, Riedinger MS, Stewart ME, Mochizuki C, Gray R. 1984. Ann Intern Med 100: 197-201
136. Harris HE, Raucci A. 2006. EMBO Rep 7: 774-8
137. Lotze MT, Tracey KJ. 2005. Nat Rev Immunol 5: 331-42
138. Kokkola R, Andersson A, Mullins G, Ostberg T, Treutiger CJ, et al. 2005. Scand J Immunol 61: 1-9
139. Bonaldi T, Talamo F, Scaffidi P, Ferrera D, Porto A, et al. 2003. EMBO J 22: 5551-60
140. Youn JH, Shin JS. 2006. J Immunol 177: 7889-97
141. Antoine DJ, Williams DP, Kipar A, Jenkins RE, Regan SL, et al. 2009. Toxicol Sci 112: 521-31
142. Ehrchen JM, Sunderkotter C, Foell D, Vogl T, Roth J. 2009. J Leukoc Biol 86: 557-66
143. Passey RJ, Xu K, Hume DA, Geczy CL. 1999. J Leukoc Biol 66: 549-56
144. Vogl T, Propper C, Hartmann M, Strey A, Strupat K, et al. 1999. J Biol Chem 274: 25291-6
145. Winningham-Major F, Staecker JL, Barger SW, Coats S, Van Eldik LJ. 1989. J Cell Biol 109: 3063-71
146. Ambartsumian N, Klingelhofer J, Grigorian M, Christensen C, Kriajevska M, et al. 2001. Oncogene 20: 4685-95
147. Heizmann CW, Fritz G, Schafer BW. 2002. Front Biosci 7: d1356-68
148. Wolf R, Lewerenz V, Buchau AS, Walz M, Ruzicka T. 2007. Exp Dermatol 16: 685-91
121
149. Elsner L, Flugge PF, Lozano J, Muppala V, Eiz-Vesper B, et al. 2010. J Cell Mol Med 14: 992-1002
150. Miller-Graziano CL, De A, Laudanski K, Herrmann T, Bandyopadhyay S. 2008. Novartis Found Symp 291: 196-208; discussion -11, 21-4
151. Pichler WJ. 2002. Curr Opin Allergy Clin Immunol 2: 301-5
152. Castrejon JL, Berry N, El-Ghaiesh S, Gerber B, Pichler WJ, et al. 2010. J Allergy Clin Immunol 125: 411-8 e4
153. Kindmark A, Jawaid A, Harbron CG, Barratt BJ, Bengtsson OF, et al. 2008. Pharmacogenomics J 8: 186-95
154. Gonzalez-Lopez MA, Martinez-Taboada VM, Gonzalez-Vela MC, Fernandez-Llaca H, Val-Bernal JF. 2008. Br J Dermatol 158: 1146-8
155. Mielke F, Schneider-Obermeyer J, Dorner T. 2008. Ann Rheum Dis 67: 1056-7
156. Ong MM, Latchoumycandane C, Boelsterli UA. 2007. Toxicol Sci 97: 205-13
157. Fujimoto K, Kumagai K, Ito K, Arakawa S, Ando Y, et al. 2009. Toxicol Pathol 37: 193-200
158. McKenzie R, Fried MW, Sallie R, Conjeevaram H, Di Bisceglie AM, et al. 1995. N Engl J Med 333: 1099-105
159. Duong Van Huyen JP, Landau A, Piketty C, Belair MF, Batisse D, et al. 2003. Am J Clin Pathol 119: 546-55
160. Kano Y, Inaoka M, Shiohara T. 2004. Arch Dermatol 140: 183-8
161. Callot V, Roujeau JC, Bagot M, Wechsler J, Chosidow O, et al. 1996. Arch Dermatol 132: 1315-21
162. Descamps V, Valance A, Edlinger C, Fillet AM, Grossin M, et al. 2001. Arch Dermatol 137: 301-4
163. Suzuki Y, Inagi R, Aono T, Yamanishi K, Shiohara T. 1998. Arch Dermatol 134: 1108-12
164. Carroll MC, Yueng-Yue KA, Esterly NB, Drolet BA. 2001. Pediatrics 108: 485-92
165. Mitani N, Aihara M, Yamakawa Y, Yamada M, Itoh N, et al. 2005. J Med Virol 75: 430-4
166. Seishima M, Yamanaka S, Fujisawa T, Tohyama M, Hashimoto K. 2006. Br J Dermatol 155: 344-9
122
167. Descamps V, Bouscarat F, Laglenne S, Aslangul E, Veber B, et al. 1997. Br J Dermatol 137: 605-8
168. Aihara M, Sugita Y, Takahashi S, Nagatani T, Arata S, et al. 2001. Br J Dermatol 144: 1231-4
169. Descamps V, Mahe E, Houhou N, Abramowitz L, Rozenberg F, et al. 2003. Br J Dermatol 148: 1032-4
170. Szyf M. 2010. Biochim Biophys Acta 1799: 750-9
171. Sekigawa I, Okada M, Ogasawara H, Kaneko H, Hishikawa T, Hashimoto H. 2003. Lupus 12: 79-85
172. Khan R, Schmidt-Mende J, Karimi M, Gogvadze V, Hassan M, et al. 2008. Exp Hematol 36: 149-57
173. Sanderson JP, Naisbitt DJ, Farrell J, Ashby CA, Tucker MJ, et al. 2007. J Immunol 178: 5533-42
174. Rhodes J. 1996. Immunol Today 17: 436-41
175. Uetrecht J. 2002. Drug Metab Rev 34: 651-65
176. Fieser IF. 1938. Am. J. Cancer 34: 87
177. Evans DC, Watt AP, Nicoll-Griffith DA, Baillie TA. 2004. Chem Res Toxicol 17: 3-16
178. Miller EC MJ. 1947. Cancer Research 7: 468–80
179. Koen YM, Hanzlik RP. 2002. Chem Res Toxicol 15: 699-706
180. Takakusa H, Masumoto H, Yukinaga H, Makino C, Nakayama S, et al. 2008. Drug Metab Dispos 36: 1770-9
181. Uetrecht J, Trager W. 2008. Drug metabolism: Chemical and enzymatic aspects
182. Mathews JM, Bend JR. 1993. J Pharmacol Exp Ther 265: 281-5
183. Nakayama S, Atsumi R, Takakusa H, Kobayashi Y, Kurihara A, et al. 2009. Drug Metab Dispos 37: 1970-7
184. Uetrecht J. 2005. Toxicology 209: 113-8
185. Naisbitt DJ. 2004. Toxicology 194: 179-96
186. Janmohamed A, Dolphin CT, Phillips IR, Shephard EA. 2001. Biochem Pharmacol 62: 777-86
123
187. Baron JM, Holler D, Schiffer R, Frankenberg S, Neis M, et al. 2001. J Invest Dermatol 116: 541-8
188. Oesch F, Fabian E, Oesch-Bartlomowicz B, Werner C, Landsiedel R. 2007. Drug Metab Rev 39: 659-98
189. Cross SE, Anderson C, Roberts MS. 1998. Br J Clin Pharmacol 46: 29-35
190. Vyas PM, Roychowdhury S, Khan FD, Prisinzano TE, Lamba J, et al. 2006. J Pharmacol Exp Ther 319: 488-96
191. Bhaiya P, Roychowdhury S, Vyas PM, Doll MA, Hein DW, Svensson CK. 2006. Toxicol Appl Pharmacol 215: 158-67
192. Kinobe RT, Parkinson OT, Mitchell DJ, Gillam EM. 2005. Chem Res Toxicol 18: 1868-75
193. Reilly TP, Lash LH, Doll MA, Hein DW, Woster PM, Svensson CK. 2000. J Invest Dermatol 114: 1164-73
194. Wolkenstein P, Tan C, Lecoeur S, Wechsler J, Garcia-Martin N, et al. 1998. Chem Biol Interact 113: 39-50
195. Svensson CK. 2009. Drug Metab Dispos 37: 247-53
196. Anderson RJ, Kudlacek PE, Clemens DL. 1998. Chem Biol Interact 109: 53-67
197. Buhl AE, Waldon DJ, Baker CA, Johnson GA. 1990. J Invest Dermatol 95: 553-7
198. Hamamoto T, Mori Y. 1989. Res Commun Chem Pathol Pharmacol 66: 33-44
199. Kudlacek PE, Anderson RJ, Liebentritt DK, Johnson GA, Huerter CJ. 1995. J Pharmacol Exp Ther 273: 582-90
200. Glatt H. 2000. Chem Biol Interact 129: 141-70
201. Negishi M, Pedersen LG, Petrotchenko E, Shevtsov S, Gorokhov A, et al. 2001. Arch Biochem Biophys 390: 149-57
202. Chapman E, Best MD, Hanson SR, Wong CH. 2004. Angew Chem Int Ed Engl 43: 3526-48
203. Wong KO, Tan AY, Lim BG, Wong KP. 1993. Biochem Pharmacol 45: 1180-2
204. Dressler WE, Appelqvist T. 2006. Food Chem Toxicol 44: 371-9
205. Uetrecht J. 2005. AAPS J 7: E914-21
206. Park JS, Choi IH, Lee DG, Han SS, Ha TY, et al. 1997. J Immunol 158: 5002-6
124
207. Chen M, Gandolfi J. 1997. Drug Metab Rev 29: 103-22
208. Furst SM, Luedke D, Gaw HH, Reich R, Gandolfi AJ. 1997. Toxicol Appl Pharmacol 143: 245-55
209. You Q, Cheng L, Reilly TP, Wegmann D, Ju C. 2006. Hepatology 44: 1421-31
210. Hoffman AM, Butt EM, Hickey NG. 1943. J. Am. Med. Assoc. 102: 2
211. Iverson S, Kautiainen A, Ip J, Uetrecht JP. 2010. Chem Res Toxicol 23: 1184-91
212. Neftel KA, Woodtly W, Schmid M, Frick PG, Fehr J. 1986. Br Med J (Clin Res Ed) 292: 721-3
213. Christie G, Breckenridge AM, Park BK. 1989. Biochem Pharmacol 38: 1451-8
214. Clarke JB, Neftel K, Kitteringham NR, Park BK. 1991. Int Arch Allergy Appl Immunol 95: 369-75
215. Maggs JL, Tingle MD, Kitteringham NR, Park BK. 1988. Biochem Pharmacol 37: 303-11
216. Clarke JB, Maggs JL, Kitteringham NR, Park BK. 1990. Int Arch Allergy Appl Immunol 91: 335-42
217. Donahue BA, McArthur JG, Spratt SK, Bohl D, Lagarde C, et al. 1999. J Gene Med 1: 31-42
218. Dieckhaus CM, Thompson CD, Roller SG, Macdonald TL. 2002. Chem Biol Interact 142: 99-117
219. Dieckhaus CM, Miller TA, Sofia RD, Macdonald TL. 2000. Drug Metab Dispos 28: 814-22
220. Donker AJ, Venuto RC, Vladutiu AO, Brentjens JR, Andres GA. 1984. Clin Immunol Immunopathol 30: 142-55
221. Tournade H, Pelletier L, Pasquier R, Vial MC, Mandet C, Druet P. 1990. J Immunol 144: 2985-91
222. Masson MJ, Uetrecht JP. 2004. Chem Res Toxicol 17: 82-94
223. Seguin B, Masson MJ, Uetrecht J. 2004. Chem Res Toxicol 17: 1299-302
224. Sayeh E, Uetrecht JP. 2001. Toxicology 163: 195-211
225. Masson MJ, Teranishi M, Shenton JM, Uetrecht JP. 2004. J Immunotoxicol 1: 79-93
226. Zhu X, Li J, Liu F, Uetrecht JP. 2011. Toxicol Sci 120: 331-8
125
227. Cribb AE, Lee BL, Trepanier LA, Spielberg SP. 1996. Adverse Drug React Toxicol Rev 15: 9-50
228. Jick H. 1982. Rev Infect Dis 4: 426-8
229. Peterson ME, Aucoin DP, Davis CA, Graves TK. 1988. Res Vet Sci 45: 1-3
230. Aucoin DP, Peterson ME, Hurvitz AI, Drayer DE, Lahita RG, et al. 1985. J Pharmacol Exp Ther 234: 13-8
231. Waldhauser L, Uetrecht J. 1996. Toxicology 114: 155-62
232. Waldhauser L, Uetrecht J. 1991. Drug Metab Dispos 19: 354-9
233. Elkayam O, Yaron M, Caspi D. 1999. Semin Arthritis Rheum 28: 392-7
234. Nassberger L, Sjoholm AG, Jonsson H, Sturfelt G, Akesson A. 1990. Clin Exp Immunol 81: 380-3
235. Shenton JM, Teranishi M, Abu-Asab MS, Yager JA, Uetrecht JP. 2003. Chem Res Toxicol 16: 1078-89
236. Lai WG, Zahid N, Uetrecht JP. 1999. J Pharmacol Exp Ther 291: 292-9
237. Riska PS, Joseph DP, Dinallo RM, Davidson WC, Keirns JJ, Hattox SE. 1999. Drug Metab Dispos 27: 1434-47
238. Shaffer CL, Morton MD, Hanzlik RP. 2001. J Am Chem Soc 123: 8502-8
239. Wen B, Chen Y, Fitch WL. 2009. Drug Metab Dispos 37: 1557-62
240. Antunes AM, Duarte MP, Santos PP, da Costa GG, Heinze TM, et al. 2008. Chem Res Toxicol 21: 1443-56
241. Antunes AM, Godinho AL, Martins IL, Justino GC, Beland FA, Marques MM. 2010. Chem Res Toxicol 23: 888-99
242. Antunes AM, Godinho AL, Martins IL, Oliveira MC, Gomes RA, et al. 2010. Chem Res Toxicol 23: 1714-25
243. Popovic M, Caswell JL, Mannargudi B, Shenton JM, Uetrecht JP. 2006. Chem Res Toxicol 19: 1205-14
244. Shenton JM, Popovic M, Chen J, Masson MJ, Uetrecht JP. 2005. Chem Res Toxicol 18: 1799-813
245. Seguin B, Boutros PC, Li X, Okey AB, Uetrecht JP. 2005. Chem Res Toxicol 18: 1193-202
246. Lopez-Garcia MP, Dansette PM, Mansuy D. 1994. Biochemistry 33: 166-75
126
247. Koenigs LL, Peter RM, Hunter AP, Haining RL, Rettie AE, et al. 1999. Biochemistry 38: 2312-9
248. Pacitto SR, Uetrecht JP, Boutros PC, Popovic M. 2007. J Immunotoxicol 4: 253-66
249. Bonierbale E, Valadon P, Pons C, Desfosses B, Dansette PM, Mansuy D. 1999. Chem Res Toxicol 12: 286-96
250. Lu W, Uetrecht JP. 2008. Drug Metab Dispos 36: 1624-36
251. Lu W, Li X, Uetrecht JP. 2008. J Immunotoxicol 5: 107-13
252. Epstein EH, Jr., Bonifas JM, Barber TC, Haynes M. 1984. J Invest Dermatol 83: 332-5
253. Lamson M, MacGregor T, Riska P, Erickson D, Maxfield P, et al. 1999. Clin Pharmacol Ther 65: 1
254. Faucette SR, Zhang TC, Moore R, Sueyoshi T, Omiecinski CJ, et al. 2007. J Pharmacol Exp Ther 320: 72-80
255. Negishi I, Motoyama N, Nakayama K, Senju S, Hatakeyama S, et al. 1995. Nature 376: 435-8
256. Fischer A, Picard C, Chemin K, Dogniaux S, le Deist F, Hivroz C. 2010. Semin Immunopathol 32: 107-16
257. Sakaguchi N, Takahashi T, Hata H, Nomura T, Tagami T, et al. 2003. Nature 426: 454-60
258. Cho HJ, Park SM, Hwang EM, Baek KE, Kim IK, et al. 2010. J Biol Chem 285: 25500-5
259. Zumbrun SD, Hoffman B, Liebermann DA. 2009. J Cell Biochem 108: 1220-31
260. Ju S, Zhu Y, Liu L, Dai S, Li C, et al. 2009. Eur J Immunol 39: 3010-8
261. Wu H, Sun YE. 2009. Sci Signal 2: pe17
262. Olivari S, Molinari M. 2007. FEBS Lett 581: 3658-64
263. Olivari S, Cali T, Salo KE, Paganetti P, Ruddock LW, Molinari M. 2006. Biochem Biophys Res Commun 349: 1278-84
264. Gallo LI, Lagadari M, Piwien-Pilipuk G, Galigniana MD. 2011. J Biol Chem 286: 30152-60
265. Li L, Lou Z, Wang L. 2011. Br J Cancer 104: 19-23
266. Fruman DA, Burakoff SJ, Bierer BE. 1994. FASEB J 8: 391-400
267. Kimura R, Morihara T, Kudo T, Kamino K, Takeda M. 2011. Neurosci Lett 487: 354-7
127
268. Liang WS, Dunckley T, Beach TG, Grover A, Mastroeni D, et al. 2008. Physiol Genomics 33: 240-56
269. Paliwal S, Shi J, Dhru U, Zhou Y, Schuger L. 2004. Biochem Biophys Res Commun 315: 1104-9
270. Fujitani M, Yamagishi S, Che YH, Hata K, Kubo T, et al. 2004. J Neurochem 91: 737-44
271. Beyeen AD, Adzemovic MZ, Ockinger J, Stridh P, Becanovic K, et al. 2010. J Immunol 185: 6883-90
272. Xu W, Presnell SR, Parrish-Novak J, Kindsvogel W, Jaspers S, et al. 2001. Proc Natl Acad Sci U S A 98: 9511-6
273. Wolf R, Mirmohammadsadegh A, Walz M, Lysa B, Tartler U, et al. 2003. FASEB J 17: 1969-71
274. Wolf R, Voscopoulos CJ, FitzGerald PC, Goldsmith P, Cataisson C, et al. 2006. J Invest Dermatol 126: 1600-8
275. Perera S, Holt MR, Mankoo BS, Gautel M. 2011. Dev Biol 351: 46-61
276. Attia RR, Sharma P, Janssen RC, Friedman JE, Deng X, et al. 2011. J Biol Chem 286: 23799-807
277. Nahle Z, Hsieh M, Pietka T, Coburn CT, Grimaldi PA, et al. 2008. J Biol Chem 283: 14317-26
278. Stevens C, Hupp TR. 2008. Autophagy 4: 531-3
279. Inbal B, Shani G, Cohen O, Kissil JL, Kimchi A. 2000. Mol Cell Biol 20: 1044-54
280. Lee JH, Rho SB, Chun T. 2005. Biotechnol Lett 27: 1011-5
281. Miranda M, Escote X, Ceperuelo-Mallafre V, Megia A, Caubet E, et al. 2010. Int J Obes (Lond) 34: 679-86
282. Camara Y, Mampel T, Armengol J, Villarroya F, Dejean L. 2009. Cell Physiol Biochem 24: 243-52
283. Flandin P, Donati Y, Barazzone-Argiroffo C, Muzzin P. 2005. FEBS Lett 579: 3411-5
284. Qiu XB, Shao YM, Miao S, Wang L. 2006. Cell Mol Life Sci 63: 2560-70
285. Suh WC, Burkholder WF, Lu CZ, Zhao X, Gottesman ME, Gross CA. 1998. Proc Natl Acad Sci U S A 95: 15223-8
286. Cullingford TE, Butler MJ, Marshall AK, Tham el L, Sugden PH, Clerk A. 2008. Biochim Biophys Acta 1783: 1229-36
128
287. Loughlin DT, Artlett CM. 2010. PLoS One 5: e11093
288. Wang J, Sarkar TR, Zhou M, Sharan S, Ritt DA, et al. 2010. Proc Natl Acad Sci U S A 107: 16131-6
289. Bowcock AM, Shannon W, Du F, Duncan J, Cao K, et al. 2001. Hum Mol Genet 10: 1793-805
290. Pedersen MB, Skov L, Menne T, Johansen JD, Olsen J. 2007. J Invest Dermatol 127: 2585-95
291. Lin DS, Yeung CY, Liu HL, Ho CS, Shu CH, et al. 2011. Am J Med Genet A 155A:
1285-9
129
Appendices Appendix 1. The complete list of genes with a significant change in gene expression in the skin 6
h after 12-OH-NVP treatment.
130
131
132
133
134
135
136
Appendix 2. The complete list of genes with a significant change in gene expression in the skin 6
h after NVP treatment.