Systems Pharmacological Analysis of Drugs Inducing Stevens

Mar 26, 2015 - Our pharmacological network analysis identified CTNNB1 [catenin (cadherin-associated protein), beta 1, 88 kDa] as a significant interme...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/crt

Systems Pharmacological Analysis of Drugs Inducing Stevens− Johnson Syndrome and Toxic Epidermal Necrolysis Junguk Hur,*,†,‡ ChunSheng Zhao,§ and Jane P. F. Bai*,§ †

Department of Basic Science, University of North Dakota, School of Medicine & Health Sciences, Grand Forks, North Dakota 58502, United States ‡ Department of Neurology, University of Michigan, Ann Arbor, Michigan 48109, United States § Office of Clinical Pharmacology, Center for Drug Evaluation and Research, U.S. Food and Drug Administration, Silver Spring, Maryland 20993, United States S Supporting Information *

ABSTRACT: Stevens−Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are serious cutaneous adverse reactions. We mined the approved labels in Drugs@FDA, identified the SJS/TEN list of 259 small molecular drugs and biologics, and conducted systems pharmacological network analyses. Pharmacological network analysis revealed that drugs with treatment-related SJS and/or TEN are pharmacologically diverse and that the largest subnetwork is associated with antiepileptic drugs and their pharmacological targets. Our pharmacological network analysis identified CTNNB1 [catenin (cadherin-associated protein), beta 1, 88 kDa] as a significant intermediator. This protein is involved in maintaining the functional integrity of the epithelium through regulating cell growth and adhesion between cells in various organs, including the skin. Leveraging a publicly accessible genome-wide transcriptional expression database, we found that human leukocyte antigen-related (HLA) genes were significantly perturbed by various SJS/TEN-inducing drugs. Notably, carbamazepine (CBZ) perturbed several HLA genes, among which HLADQB1*0201 was reportedly shown to be associated with CBZ-induced SJS/TEN in caucasians. In short, systems analysis by leveraging a publicly accessible knowledge base and databases could produce meaningful results for further mechanistic investigation. Our study sheds light on the utility of systems pharmacology analysis for gaining insight into clinical drug toxicity.

■

INTRODUCTION Major drug-induced serious cutaneous adverse reactions (SCAR) include Stevens−Johnson syndrome (SJS), toxic epidermal necrolysis (TEN), and drug reaction with eosinophilia and systemic symptoms (DRESS). SJS and TEN result in high mortality.1 With SJS and TEN as the keywords, 482 publications were found in PubMed, and 86 publications were found with a combination of SJS, TEN, and genetic. Only 22 publications were found if genetic was replaced with mechanism. Clearly, a limited number of studies have been devoted to investigating drug-induced SJS/TEN at the molecular levels. Over past decades, major progress has been made in identifying the alleles associated with treatment-emergent SCARs. Specifically, HLA alleles have been identified as markers for predicting the risk for DRESS,2 SJS,3 or SJS/ TEN4 in specific ethnic groups, revealing the prominent role of the immune system in these conditions. Different HLA allele markers identified for a specific drug-induced SCAR have been attributed to different clinical diagnostic criteria used and/or to different ethnicities.4,5 In this study, we proposed to mine the FDA-approved labels in the Drugs@FDA database, to compile all of the drugs with SJS and TEN described in approved labels, and to conduct network analysis to highlight the pharmaco© XXXX American Chemical Society

logical network and uncover the significantly differentially expressed genes. In addition to the major histocompatibility complex-related genes, we identified some genes that warrant further investigation in order to gain insight into the mechanism of drug-induced SJS/TEN.

■

EXPERIMENTAL PROCEDURES

Compilation and Curation of SJS/TEN List. To understand the diversity in pharmacological characteristics of the drugs that have been shown to cause treatment-related SJS/TEN, we performed our systems pharmacology approach in the same manner as that previously applied to the studies of drug-induced peripheral neuropathy and rhabdomyolysis.6,7 We conducted text mining of all approved labels in the Drugs@FDA database8 and curated the list of drugs for which labels include SJS/TEN as treatment-related adverse reactions observed during clinical trials or in the postapproval setting. The list includes both small-molecule drugs and biologics. Construction of Pharmacological Network. To better understand how these drugs from different classes are potentially interconnected, we constructed the pharmacological network of drugs-induced SJS/TEN using our previously published method.6,7 Briefly, a pharmacological network, including drugs and their known Received: December 18, 2014

A

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology

Figure 1. Pharmacological network of SJS/TEN-inducing drugs. (A) The base pharmacological network includes 258 drugs with treatmentassociated SJS and TEN and their nominal human targets. The layout of the network was generated using edge-embedded spring layout available in Cytoscape (http://cytoscape.org/). Node color represents the community structures identified by the Greedy algorithm implemented in GLay (http://brainarray.mbni.med.umich.edu/glay/), where gray is used as a default color and nongray colors were used for the 12 biggest subnetworks. (B) The biggest subnetwork with 25 drugs and 55 targets is highlighted in yellow. Drugs were mostly from the antiepileptic class. (C) The second biggest subnetwork includes 24 drugs and 35 targets. Drugs were mostly antidepressive and antihypertensive agents. (D) The third biggest subnetwork (in blue) includes diverse drug classes such as antirheumatic, immunosuppressive, and antineoplastic agents, whereas the fourth biggest subnetwork (in cyan) contains mostly antipain agents. Node shape represents the type of entity: rectangle for drug and circle for drug target. Node size corresponds to the number of interacting nodes. targets, was generated in Cytoscape,9 an open-source network visualization and analysis tool based on the drug−target association collected from DrugBank10 and Therapeutic Target Database (TTD).11 Identification of Significant Intermediators. To identify potential common but crucial players in the pathogenesis of diverse drug-induced SJS/TEN, we extended the pharmacological network by adding the proteins or genes, denoted as intermediators, that have protein−protein interactions (PPIs) or genetic interactions (GIs) with the known targets of SJS/TEN-inducing drugs. The PPIs and GIs were derived from the BioGRID, release 3.2.114 (http://thebiogrid.org/),12 and 3371 intermediators were found to be indirectly interacting with at least one drug that induces SJS/TEN. In order to identify the intermediators significantly associated with SJS/TEN in the pharmacological network, the drug-degree of each intermediator, defined as the number of drugs indirectly interacting via the known targets, was compared against 1000 random extended networks using a Z-test. Significant intermediators were defined as those that have at least five interacting drugs, a Benjamini−Hochberg corrected P-value < 0.05 for the Z-test, and a minimum of a 1.5-fold enrichment of the drug-degree in the SJS/TEN network compared to the average drugdegrees of random networks. Differential Gene Expression Perturbed by SJS/TEN-Inducing Drugs. Transcriptomic profiles of the SJS/TEN-inducing drugs were obtained from the Connectivity Map (CMap) generated by the Broad Institute of MIT and Harvard (http://www.broadinstitute.org/ cmap/).13 The CMap contains a collection of genome-wide transcriptional expression data from cultured human cancer cells treated with bioactive small-molecules profiled using Affymetrix Human Genome U133A 2.0 arrays. In total, 123 curated SJS/TEN-inducing drugs were

found to be included in CMap build 02. In order to create a representative gene signature for each drug, the 250 most upregulated genes, 250 most downregulated genes, and gene expression profiles from different cell types, dosages, and treatment durations were combined using the Kru-Bor merging method.14 A simple frequencybased approach was used to identify the most frequently perturbed genes by the SJS/TEN-inducing drugs. The genes in the signatures of 123 drugs were ranked by frequency; the top 5% most frequently perturbed genes were further examined for their enriched biological functions using Database for Annotation, Visualization and Integrated Discovery, v6.7 (DAVID; http://david.abcc.ncifcrf.gov/).15,16 In particular, the HLA genes were closely examined for their perturbation by SJS/TEN-inducing drugs.

■

RESULTS

Compilation and Curation of SJS/TEN List. Text mining and curation of all approved labels in the Drugs@FDA database8 resulted in identification of 259 drugs having SJS/ TEN as treatment-related adverse reactions observed during clinical trials or in the postapproval setting (see Table S1 for the complete list). Of interest, the biologic products identified were interferon alfa-2B, aldesleukin (a human recombinant interleukin-2 product), interferon alfacon-1, rituximab, infliximab, etanercept, interferon gamma-1B, peginterferon alfa-2B, peginterferon alfa-2A, ibritumomab tiuxetan, adalimumab, efalizumab, certolizumab pegol, and ipilimumab. Pharmacological Network of SJS/TEN-Inducing Drugs. There were a total of 245 small-molecule drugs and 14 B

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology Table 1. Top 20 Most Significant Intermediatorsa no. of interacting drugs (drug-degree) gene ID

symbol

10987 1499 8454 857 2033 7415 4734

COPS5 CTNNB1 CUL1 CAV1 EP300 VCP NEDD4

8451 5705 26270 142 5925 9978 5371 79139 5740 9844 55829 10659 4924

CUL4A PSMC5 FBXO6 PARP1 RB1 RBX1 PML DERL1 PTGIS ELMO1 SELS CELF2 NUCB1

a

description COP9 signalosome subunit 5 catenin (cadherin-associated protein), beta 1, 88 kDa cullin 1 caveolin 1, caveolae protein, 22 kDa E1A binding protein p300 valosin containing protein neural precursor cell expressed, developmentally downregulated 4, E3 ubiquitin protein ligase cullin 4A proteasome (prosome, macropain) 26S subunit, ATPase, 5 F-box protein 6 poly(ADP-ribose) polymerase 1 retinoblastoma 1 ring-box 1, E3 ubiquitin protein ligase promyelocytic leukemia derlin 1 prostaglandin I2 (prostacyclin) synthase engulfment and cell motility 1 VCP-interacting membrane protein CUGBP, Elav-like family member 2 nucleobindin 1

SJS/TEN network

1000 random networks

53 49 45 44 43 41 38

29.2 23.9 24.5 25.3 28.3 23.8 24.1

3.40 6.82 8.54 8.37 1.17 1.47 1.11

× × × × × × ×

10−5 10−7 10−5 10−4 10−2 10−3 10−2

1.8 2.1 1.8 1.7 1.5 1.7 1.6

36 31 31 30 29 28 28 27 24 24 24 24 24

11.4 14.8 19.7 18.1 13.5 7.2 16.1 8.9 5.0 5.4 5.6 5.6 6.0

3.90 1.22 2.15 1.30 1.98

× × × × ×

10−13 10−4 10−2 10−2 10−2

3.1 2.1 1.6 1.7 2.2 3.9 1.7 3.0 4.8 4.5 4.3 4.3 4.0

FDR

enrichment fold

1.06 × 10−2 1.65 × 10−9 9.20 × 10−8

4.95 × 10−13

TF role cofactor cofactor

cofactor

cofactor cofactor cofactor cofactor

FDR, false discovery rate; TF, transcription factor.

Table 2. Significantly Over-represented Biological Functions among the 612 Frequently Perturbed Genes by SJS/TEN-Inducing Drugsa cluster

GO ID

GO term

count

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

GO:0042981 GO:0006357 GO:0016568 GO:0010605 GO:0006796 GO:0009967 GO:0040029 GO:0007010 GO:0007049 GO:0002474 GO:0032583 GO:0002520 GO:0030518 GO:0045637 GO:0065003

regulation of apoptosis regulation of transcription from RNA polymerase II promoter chromatin modification negative regulation of macromolecule metabolic process phosphate metabolic process positive regulation of signal transduction regulation of gene expression, epigenetic cytoskeleton organization cell cycle antigen processing and presentation of peptide antigen via MHC class I regulation of gene-specific transcription immune system development steroid hormone receptor signaling pathway regulation of myeloid cell differentiation macromolecular complex assembly

60 55 25 52 58 24 13 31 56 6 14 22 9 10 42

P-value 4.64 1.02 9.92 1.35 4.67 7.29 2.64 9.83 3.37 3.06 1.53 1.65 1.34 1.14 1.15

× × × × × × × × × × × × × × ×

10−7 10−6 10−5 10−5 10−4 10−4 10−5 10−4 10−6 10−4 10−3 10−3 10−3 10−3 10−3

FE

BH P-value

1.99 2.02 2.43 1.89 1.59 2.17 4.5 1.9 1.93 9.42 2.79 2.13 4.14 3.81 1.69

1.32 5.81 8.52 3.49 2.58 3.64 3.95 4.63 1.06 1.83 6.05 6.33 5.53 5.02 4.90

× × × × × × × × × × × × × × ×

10−3 10−4 10−3 10−3 10−2 10−2 10−3 10−2 10−3 10−2 10−2 10−2 10−2 10−2 10−2

a

GO, gene ontology, FE, fold enrichment (the relative ratio of corresponding gene proportions in the SJS/TEN set vs background (complete human genome)); BH, Benjamini−Hochberg.

subnetworks are indicated by nongray node colors. Each subnetwork was enriched with specific classes of drugs: the largest subnetwork with antiepileptic agents (Figure 1B), the second largest subnetwork, with antidepressive and antihypertensive agents (Figure 1C), the third largest subnetwork, with antirheumatic, immunosuppressive, and antineoplastic agents (blue nodes in Figure 1D), and the fourth largest subnetwork, with NSAIDs (cyan nodes in Figure 1D). Identification of Significant Intermediators. We extended the base pharmacological network by referencing the BioGRID database12 to add intermediators that had either GIs or PPIs with the known targets of SJS/TEN-inducing drugs. A total of 3371 intermediators were found to be indirectly

biologics with SJS and/or TEN listed as treatment-related adverse reactions (Table S1). Among these 259 drugs, 234 had at least one target, 193 of which included human proteins as their target. Figure 1A illustrates the complete pharmacological network of SJS/TEN, including 234 drugs and human protein targets (see the Cytoscape network file available as Supporting Information). The pharmacological network included multiple subnetworks, within which drugs and targets are highly connected. The fast Greedy community structure analysis algorithm,17 an automatic method for identifying highly interconnected subnetworks based on the network topology, was applied to the SJS/TEN pharmacological network. The top 12 largest C

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

a

D

HLA-B

HLA-G

HLA-A

HLA-E

HLA-DPA1

HLA-DQB1

HLA-DRB1

HLA-DMB

HLA-F

HMHA1 HLA-DOA

HLA-DMA

HLA-DRB6

CD74

HLA-DOB

HLA-DRB3

CIITA

HLA-K

HLA-DQA1

HLA-DQB2

HLA-DRA

3106

3135

3105

3133

3113

3119

3123

3109

3134

23526 3111

3108

3128

972

3112

3125

4261

3138

3117

3120

3122

major histocompatibility complex, class I, B major histocompatibility complex, class I, G major histocompatibility complex, class I, A major histocompatibility complex, class I, E major histocompatibility complex, class II, DP alpha 1 major histocompatibility complex, class II, DQ beta 1 major histocompatibility complex, class II, DR beta 1 major histocompatibility complex, class II, DM beta major histocompatibility complex, class I, F histocompatibility (minor) HA-1 major histocompatibility complex, class II, DO alpha major histocompatibility complex, class II, DM alpha major histocompatibility complex, class II, DR beta 6 (pseudogene) CD74 molecule, major histocompatibility complex, class II invariant chain major histocompatibility complex, class II, DO beta major histocompatibility complex, class II, DR beta 3 class II, major histocompatibility complex, transactivator major histocompatibility complex, class I, K (pseudogene) major histocompatibility complex, class II, DQ alpha 1 major histocompatibility complex, class II, DQ beta 2 major histocompatibility complex, class II, DR alpha

major histocompatibility complex, class I-related

description

1

1

1

1

1

2

2

2

3

3

5 5

7

10

11

12

12

12

12

16

18

23

no. of drugs drugs

Iopromide

Metolazone

Fluvastatin

Repaglinide

Verapamil

Iopromide; Lovastatin

Meropenem; Trimethoprim

Bupropion; Trimethoprim

Aciclovir; Fenoprofen; Rifampicin

Acetaminophen; Omeprazole; Valdecoxib

Acetazolamide; Brinzolamide; Cefepime; Indapamide; Ofloxacin Carbamazepine; Cefazolin; Gefitinib; Indapamide; Saquinavir

Cefotaxime; Gefitinib; Hydrochlorothiazide; Minocycline; Repaglinide; Saquinavir; Simvastatin

Albendazole; Ethosuximide; Famotidine; Levobunolol; Naproxen; Piperacillin; Piroxicam; Topiramate; Valdecoxib; Vancomycin

Cefuroxime; Ciprofloxacin; Clindamycin; Etoposide; Famotidine; Hydrochlorothiazide; Ketorolac; Lincomycin; Ofloxacin; Simvastatin; Thalidomide; Tolmetin Aztreonam; Cefoxitin; Cephalexin; Chlorambucil; Cinoxacin; Diltiazem; Ethosuximide; Fenoprofen; Fluoxetine; Hydrochlorothiazide; Ketoprofen; Valdecoxib Amoxicillin; Carbamazepine; Cefuroxime; Fenoprofen; Ketoprofen; Metronidazole; Nifedipine; Oxaprozin; Pentamidine; Rifampicin; Simvastatin; Tolmetin Captopril; Cefepime; Celecoxib; Ciprofloxacin; Clindamycin; Diclofenac; Diltiazem; Gefitinib; Hydrochlorothiazide; Metronidazole; Tamoxifen

Aciclovir; Bupropion; Captopril; Carbamazepine; Clozapine; Diclofenac; Etoposide; Gefitinib; Hydrochlorothiazide; Ifosfamide; Iopromide; Lincomycin; Mefenamic Acid; Minocycline; Norfloxacin; Piperacillin; Piroxicam; Repaglinide; Rifampicin; Rofecoxib; Sulfasalazine; Vancomycin; Zidovudine Amoxicillin; Brinzolamide; Cefadroxil; Diltiazem; Doxycycline; Erythromycin; Etoposide; Furosemide; Ifosfamide; Lomefloxacin; Metolazone; Minocycline; Omeprazole; Piroxicam; Sulindac; Tamoxifen; Vancomycin; Zidovudine Bumetanide; Captopril; Carbamazepine; Cefadroxil; Cefazolin; Ciprofloxacin; Erythromycin; Glimepiride; Ifosfamide; Ketorolac; Lisinopril; Meropenem; Omeprazole; Pyrimethamine; Rifampicin; Sulfamethoxazole Brinzolamide; Bupropion; Cefoxitin; Diflunisal; Erythromycin; Etoposide; Ifosfamide; Ketorolac; Leflunomide; Minocycline; Topiramate; Zidovudine

MHC, major histocompatibility complex; HLA, human leukocyte antigen.

MR1

symbol

3140

gene ID

Table 3. Major Histocompatibility Complex-Related Genesa

Chemical Research in Toxicology Article

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology Table 4. Gene Overlap Matrixa symbol

description

DEGs

MAP4 FGFR1 BAZ1B VEGFA TYMS SQLE SLC7A11 RXRA RARA RANBP1 PDCD6 NR3C1 MR1 HLA-G HLA-B INSIG1 FKBP1A ESR1 DRD2 DHFR CACNB1 BCL2L1 PDPK1 DHODH

microtubule-associated protein 4 fibroblast growth factor receptor 1 bromodomain adjacent to zinc finger domain, 1B vascular endothelial growth factor A thymidylate synthetase squalene epoxidase solute carrier family 7 (anionic amino acid transporter light chain, xc-system), member 11 retinoid X receptor, alpha retinoic acid receptor, alpha RAN binding protein 1 programmed cell death 6 nuclear receptor subfamily 3, group C, member 1 (glucocorticoid receptor) major histocompatibility complex, class I-related major histocompatibility complex, class I, G major histocompatibility complex, class I, B insulin induced gene 1 FK506 binding protein 1A, 12 kDa estrogen receptor 1 dopamine receptor D2 dihydrofolate reductase calcium channel, voltage-dependent, beta 1 subunit BCL2-like 1 3-phosphoinositide dependent protein kinase-1 dihydroorotate dehydrogenase (quinone)

× × × × × × × × × × × × × × × × × × × × × × × ×

drug target

HLA genes

significant intermediators

× × × × × × × × × × × × × × × × × × × × × × × ×

a × indicates that the corresponding gene was included in the respective gene set. DEGs, differentially expressed genes (defined by the top/bottom 5% of genes that are most frequently perturbed by SJS/TEN-inducing drugs; minimum five drugs).

number of drugs by which the fifth percentile gene was perturbed) were selected for functional enrichment analysis using DAVID. Table 2 lists the representative terms of the top 15 clusters of significantly over-represented GO terms in the 612 frequently perturbed genes by SJS/TEN inducing drugs. These functions included diverse biological functions such as regulation of apoptosis, cytoskeleton organization, antigen processing, and presentation of peptide antigen via MHC class I. By focusing on the human leukocyte antigen-related genes that were differentially perturbed by the SJS/TEN-inducing drugs, we found that many human leukocyte antigen (HLA) genes were perturbed by at least 10 SJS/TEN-inducing drugs (Tables 3 and S6). These genes [MR1 (HLA-LS), HLA-B, HLA-G, HLA-A, HLA-DQB1, HLA-E, HLA-DRB1, and HLADMB] encode major histocompatibility complex (MHC) molecules on the cell surface of antigen-presenting cells. We further examined if any of these frequently perturbed genes included know targets of SJS/TEN-inducing drugs, significant intermediators from the pharmacological network, or HLA genes. Figure S1 illustrates the overlap among these four gene sets. Among 612 DEGs, 24 were also shared by other three gene sets, of which two-thirds were known drug targets (Table 4).

interacting with at least one SJS/TEN-inducing drug. The drugdegree, defined as the number of drugs indirectly interacting via the known targets, of each intermediator was compared against those obtained in the extended networks of the 1000 randomly generated pharmacological networks (equal in size with the SJS/TEN network). Among the intermediators, 168 (Table S2) met our significance criteria: having at least five interacting drugs, a Benjamini−Hochberg corrected P-value < 0.05, and a minimum of 1.5-fold enrichment of drug-degree in the SJS/ TEN network compared to the average drug-degrees of the random networks. A total of 61 biological functions (in terms of SwissProt keywords, Gene Ontology terms, and KEGG pathways) were found to be enriched among these significant intermediators, including acetylation, phosphoproteins, and chromatic organization (Table S3). The 20 significant intermediators with the highest number of drug-degrees are listed in Table 1, which included seven transcription cofactors according to MatBase (Genomatix, Munich, Germany; http://www.genomatix.de/). Among the top 20 highly connected intermediators, 10 intermediators having an enrichment fold equal to or greater than 2 are as listed in Table S4 along with summaries of their main biological functions. Differential Gene Expression Perturbed by SJS/TENInducing Drugs. Among the SJS/TEN-inducing drugs, there were 123 that were profiled and included in the Connectivity Map database. Differentially expressed genes (DEGs) were defined to be the top 250 most upregulated and 250 most downregulated by each drug treatment (combined by the KruBor merging method per drug). A total of 10 434 genes were found to be a DEG to at least one SJS/TEN-inducing drug (Table S5), and 612 genes perturbed by at least 15 drugs (the

■

DISCUSSION

Clinical phenotypes of SJS/TEN include skin detachment, epidermal necrolysis, systemic involvement (including intestine and lung), and erosion of mucous membranes.5 Treatmentrelated SJS/TEN is associated with drugs across therapeutic areas and pharmacological classes. E

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology

function and survival. From the perspective of cellular function, the protein encoded by COPS8 (COP9 signalosome subunit 8) is similar to the 19S unit of the 26S proteasome (http://www. ncbi.nlm.nih.gov/gene/). The top 20 intermediators listed in Table 1 with an enrichment factor being equal to or greater than 2 are involved in essential cellular proliferation, survival, and growth, whereas nucleobindin 1, VCP-interacting membrane protein (encoded by the VIMP gene), and proteasome 26S subunit, ATPase 5 (encoded by the PSMC5 gene) are involved in immunological responses. Considering the clinical phenotypes, the epithelium is the tissue to which the sensitized T cells are homing. In addition to the skin, many organs in the human body are lined with epithelial cells. Epithelial cells line the gastrointestinal tract, pulmonary alveoli, trachea, nasal cavity, cornea, bladder, kidney, and so on. It seems reasonable that CTNNB1 [catenin (cadherin-associated protein), beta 1, 88 kDa] is a significant intermediator identified in our pharmacological network analysis, considering that the protein encoded by this gene is involved in maintaining epithelial cell layers by regulating cell growth and adhesion between cells. The role of CTNNB1 in the pathogenesis of drug-induced SJS/TEN is unknown due to a lack of published reports. This gene, however, seems to play a role in re-epitheliazation and wound healing. Myofibroblasts were reportedly involved in re-epithelialization in a patient suffering from allopurinol-induced SJS/TEN.24 The transition of fibroblasts to myofibroblasts is essential for wound healing, and beta catenin is involved in this myfibroblast transition.25 Mechanistic involvement of T cell receptors, major histocompatibility complex (MHC) class molecules, and antigen presentation processes have been demonstrated at the cellular and molecular levels for drug-induced skin hypersensitivity,22 especially in CBZ-induced SJS and TEN.26 Specific T cell clones that expressed VA-22-FISGTY/VB-11ISGSY were shown to play a key role in CBZ-induced SJS/ TEN,26 and this observation provided insight into the tolerance to CBZ seen in HLA-B*1502 carriers. In short, these focused studies reveal that interactions among the immune system, patient biology, and drugs that underlie the onset of SJS and TEN are very complex. Utilizing the genome-wide transcriptional expression data of Connectivity Map in which cancer cell lines were used, we were able to find that CBZ perturbed MR1, HLA-G, HLA-DBQ1, and HLA-DOA. Among these genes, mutations of both HLA-DBQ1 and HLA-DQA were previously found to be associated with CBZ-induced SJS/TEN observed in caucasians.18 Our new findings regarding perturbation of MR1 and HLA-G by CBZ warrant studies to understand their role in drug-induced SJS/TEN. For CBZ-induced SJS/TEN, HLA A*0101, Cw*0701, B*0801, DRB1*0301, DQA1*0501, DQB1*0201 were identified for caucasians,18 and HLAA*3101, for northern Europeans27 as well as Japanese.28 Although we expect that cancer cell lines will behave differently from immune cells in humans, it is interesting that our analysis was able to identify HLA-DBQ1 for possible linkage to CBZinduced SJS/TEN. Moreover, a functional enrichment analysis of differentially perturbed genes identified the role of immune genes in drug-induced SJS/TEN even though we utilized the data from cancer cell lines. The cancer cell lines used for generating the Connectivity Map data are likely derived from caucasians and may be the underlying reason for our observed perturbation of HLA-DBQ1. Among the top/bottom 5% most frequently perturbed genes, our analysis was able to show the discrete overlap of these

Although the majority of the SJS/TEN-inducing drugs are a small-molecule in nature, there are 14 biological products with SJS/TEN noted in their approved labels, which include interferon alfa-2B, aldesleukin (a human recombinant interleukin-2 product), interferon alfacon-1, rituximab, infliximab, etanercept, interferon gamma-1B, peginterferon alfa-2B, peginterferon alfa-2A, ibritumomab tiuxetan, adalimumab, efalizumab, certolizumab pegol, and ipilimumab. In terms of biological products, each and every one of them is designed to have specific targets or to be recombinant cytokines or pegylated recombinant cytokines. Interestingly, among the aforementioned biological products, rituximab and ibritumomab tiuxetan are antibodies directed against CD-20, efalizumab is an antibody against CD11a, and ipilimumab is an antibody against cytotoxic T-lymphocyte antigen 4 (CTLA-4), whereas infliximab, etanercept, adalimumab, and certolizumab pegol are tumor necrosis factor blockers. The remaining drugs are either cytokines or pegylated cytokines, which are small proteins involved in communication between immune cells and in cell signaling. These results suggest that biological products, related to immunological actions, seem to have the potential to cause treatment-related SJS and TEN. Antiepileptic drugs constituted the largest subnetwork in our pharmacological network. By leveraging PubMed, drug-induced SJS/TEN has been indeed extensively studied in antiepileptic drugs, including carbamazepine (CBZ), oxcarbazepine, and phenytoin.18 Among these drugs, CBZ has been most extensively studied to unravel the interplay among the drug, patient biology, and immune system in drug-induced SJS/TEN. Different genomic biomarkers have been associated with CBZinduced SJS/TEN, notably HLA-B*1502 is a genomic markers for Han Chinese3,19 and for northern Indians.20 In addition to specific genetic carriers, interactions between drug molecules and the immune system have been studied.21,22 Nonetheless, HLA alleles are most prominently shown to be of statistical significance. HLA-B*1502 is consistently identified as a genomic biomarker for SJS/TEN induced by carbamazepine, phenytoin, lamotrigine, and oxcarbazepine.18 There are, however, Han Chinese who are not HLA-B*1502 carriers but develop CBZ-induced SJS/TEN or who are HLA-B*1502 carriers but are tolerant to CBZ-induced SJS/TEN,23 indicating the complexity of immune regulation at the gene level. In the pharmacological network analysis, no linkage was established between drug targets and HLA proteins, indicating the lack of biological knowledge and reflecting the limited number of mechanistic studies published. From the top 20 intermediators that interact with SJS/TENinducing drugs (Table 1), COP9 signalosome subunit 5 (encoded by COPS5) was the mostly highly connected significant intermediator, interacting with 53 SJS/TENinducing drugs and showing about 1.8-fold enrichment compared to the average drug-degree of 29.2 drugs in the 1000 random networks. Among the significant intermediators, we found that phosphorylation and acetylation were the two most significantly over-represented biological functions in terms of Gene Ontology. Phosphorylation and acetylation are involved in post-translational modification of proteins, indicating that signaling pathways or networks are connected to our pharmacological network. Other significant terms were related to DNA repair and the ubiquitin conjugation pathway. We also noted that the ubiquitin-mediated proteolysis pathway was most significantly enriched. These results are not surprising, as these terms are related to general cellular F

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology

org/) network file. This material is available free of charge via the Internet at http://pubs.acs.org.

genes with pharmacological targets, with HLAs, and with intermediators, as illustrated in Table 4. These results indicate the importance of using multiple approaches for integrating gene expression, pharmacological targets, and intermediators for shedding light on drug toxicity within the context of systems biology. As sciences advance, it will be possible to evaluate a more extensive linkage among the biological layers of gene expression, pharmacological targets, and intermediators for defining the biological cube of specific drug-induced adverse reactions. In terms of mechanisms, both hapten and p-i concepts have been proposed for drug-induced hypersensitivity.22 The former require a covalent modification of a protein by a drug molecule that results in immune recognition of the modified protein as an antigen and thereby induces immune responses, whereas the latter involve direction interaction between drug molecules and T cell receptors or MHC molecules that results in elicitation of immune responses. Interestingly, at the molecular level, the T cell clone expressing VA-22-FISGTY/VB-11-ISGSY was able to recognize the CBZ−HLA-B*1502 complex, resulting in CBZinduced skin hypersensitivity.26 Interaction between CBZ and HLA-B*1502 seems to be noncovalent in nature.29 Subsequently, computer-aided modeling was used to demonstrate multiple possible direct interactions between the 5-carboxamide group of CBZ and the pockets of HLA-B*1502.30 In light of these reports, we attempted a structural analysis to determine whether there is any correlation between chemical structures of small-molecule drugs and treatment-related SJS/TEN. Due to the diversity in the pharmacological classes of SJS/TENinducing small-molecule drugs, no correlation was observed by utilizing chemical structural similarity analysis with Tanimoto coefficients. Considering that noncovalent and covalent interactions are both possible, chemical structure-based analysis might not be insightful. On the other hand, pharmacological network analysis along with analysis of differential gene expression profiles identified the HLA-related and cellular function-related genes that could potentially be involved in drug-induced SJS/TEN and provided mechanistic insight into such adverse reactions. This study illustrates the utility of a combined analysis of pharmacological network and drug-perturbed transcriptomic profiles. Such integrated analysis enables us to reference the biological functions of the genes that are perturbed by SJS/ TEN-inducing drugs for depicting the underlying biological network. Most interestingly, our finding of the role of HLA genes seems to be consistent with the literature on HLA genomic biomarkers for drug-induced SJS/TEN.

■

■

AUTHOR INFORMATION

Corresponding Authors

*(J.H.) Phone: 701-777-6814. E-mail: [email protected]. edu. *(J.P.F.B.) Phone: 301-796-2473. E-mail: [email protected]. Funding

C.Z. was supported as an ORISE through a Critical Path Grant from the Center for Drug Evaluation and Research to J.P.F.B. and by a Postdoctoral Fellowship from the Juvenile Diabetes Research Foundation to J.H. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank Dr. Darrell Abernethy (Office of Clinical Pharmacology, U.S. Food and Drug Administration) for his comments and suggestions on the manuscript. The authors also thank Dr. Zhichao Liu at the National Center of Toxicological Research, U.S. Food and Drug Administration, for a structural similarity analysis among the SJS/TEN-inducing drugs. The views expressed are those of the authors and do not necessarily represent the position of, nor imply endorsement from, the U.S. Food and Drug Administration or the U.S. government.

■ ■

ABBREVIATIONS SJS, Stevens−Johnson syndrome; TEN, toxic epidermal necrolysis REFERENCES

(1) Mockenhaupt, M. (2014) Stevens−Johnson syndrome and toxic epidermal necrolysis: clinical patterns, diagnostic considerations, etiology, and therapeutic management. Semin. Cutaneous Med. Surg. 33, 10−16. (2) Goncalo, M., Coutinho, I., Teixeira, V., Gameiro, A. R., Brites, M. M., Nunes, R., and Martinho, A. (2013) HLA-B*58:01 is a risk factor for allopurinol-induced DRESS and Stevens−Johnson syndrome/toxic epidermal necrolysis in a Portuguese population. Br. J. Dermatol. 169, 660−665. (3) Chung, W. H., Hung, S. I., Hong, H. S., Hsih, M. S., Yang, L. C., Ho, H. C., Wu, J. Y., and Chen, Y. T. (2004) Medical genetics: a marker for Stevens−Johnson syndrome. Nature 428, 486. (4) Genin, E., Chen, D. P., Hung, S. I., Sekula, P., Schumacher, M., Chang, P. Y., Tsai, S. H., Wu, T. L., Bellon, T., Tamouza, R., Fortier, C., Toubert, A., Charron, D., Hovnanian, A., Wolkenstein, P., Chung, W. H., Mockenhaupt, M., and Roujeau, J. C. (2014) HLA-A*31:01 and different types of carbamazepine-induced severe cutaneous adverse reactions: an international study and meta-analysis. Pharmacogenomics J. 14, 281−288. (5) Pirmohamed, M., Friedmann, P. S., Molokhia, M., Loke, Y. K., Smith, C., Phillips, E., La Grenade, L., Carleton, B., Papaluca-Amati, M., Demoly, P., and Shear, N. H. (2011) Phenotype standardization for immune-mediated drug-induced skin injury. Clin. Pharmacol. Ther. 89, 896−901. (6) Hur, J., Guo, A. Y., Loh, W. Y., Feldman, E. L., and Bai, J. P. F. (2014) Integrated systems pharmacology analysis of clinical druginduced peripheral neuropathy. CPT: Pharmacometrics Syst. Pharmacol. 3, e114. (7) Hur, J., Liu, Z., Tong, W., Laaksonen, R., and Bai, J. P. (2014) Drug-induced rhabdomyolysis: from systems pharmacology analysis to biochemical flux. Chem. Res. Toxicol. 27, 421−32. (8) Drugs@ FDA. http://www.accessdata.fda.gov/scripts/cder/ drugsatfda/index.cfm (accessed July, 2014).

ASSOCIATED CONTENT

* Supporting Information S

Figure S1: Venn diagram of DEGs, intermediators, HLA genes, and drug targets. Table S1: 259 drugs having SJS/TEN as treatment-related adverse reactions observed during clinical trials or in the postapproval setting. Table S2: 168 intermediators meeting the significance criteria. Table S3: 61 biological functions found to be enriched among the significant intermediators. Table S4: 10 intermediators having an enrichment fold equal to or greater than 2 along with a summary of their main biological functions. Table S5: 10 434 genes found to be a DEG to at least one SJS/TEN-inducing drug. Table S6: Human leukocyte antigen (HLA) genes perturbed by at least 10 SJS/TEN-inducing drugs. The base pharmacological network is available as a Cytoscape (http://www.cytoscape. G

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX

Article

Chemical Research in Toxicology (9) Kohl, M., Wiese, S., and Warscheid, B. (2011) Cytoscape: software for visualization and analysis of biological networks. Methods Mol. Biol. 696, 291−303. (10) Knox, C., Law, V., Jewison, T., Liu, P., Ly, S., Frolkis, A., Pon, A., Banco, K., Mak, C., Neveu, V., Djoumbou, Y., Eisner, R., Guo, A. C., and Wishart, D. S. (2011) DrugBank 3.0: a comprehensive resource for ‘omics’ research on drugs. Nucleic Acids Res. 39, D1035−1041. (11) Qin, C., Zhang, C., Zhu, F., Xu, F., Chen, S. Y., Zhang, P., Li, Y. H., Yang, S. Y., Wei, Y. Q., Tao, L., and Chen, Y. Z. (2014) Therapeutic target database update 2014: a resource for targeted therapeutics. Nucleic Acids Res. 42, D1118−1123. (12) Chatr-Aryamontri, A., Breitkreutz, B. J., Heinicke, S., Boucher, L., Winter, A., Stark, C., Nixon, J., Ramage, L., Kolas, N., O’Donnell, L., Reguly, T., Breitkreutz, A., Sellam, A., Chen, D., Chang, C., Rust, J., Livstone, M., Oughtred, R., Dolinski, K., and Tyers, M. (2013) The BioGRID interaction database: 2013 update. Nucleic Acids Res. 41, D816−823. (13) Lamb, J., Crawford, E. D., Peck, D., Modell, J. W., Blat, I. C., Wrobel, M. J., Lerner, J., Brunet, J. P., Subramanian, A., Ross, K. N., Reich, M., Hieronymus, H., Wei, G., Armstrong, S. A., Haggarty, S. J., Clemons, P. A., Wei, R., Carr, S. A., Lander, E. S., and Golub, T. R. (2006) The Connectivity Map: using gene-expression signatures to connect small molecules, genes, and disease. Science 313, 1929−1935. (14) Iorio, F., Bosotti, R., Scacheri, E., Belcastro, V., Mithbaokar, P., Ferriero, R., Murino, L., Tagliaferri, R., Brunetti-Pierri, N., Isacchi, A., and di Bernardo, D. (2010) Discovery of drug mode of action and drug repositioning from transcriptional responses. Proc. Natl. Acad. Sci. U.S.A. 107, 14621−14626. (15) Huang da, W., Sherman, B. T., and Lempicki, R. A. (2009) Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44−57. (16) Huang da, W., Sherman, B. T., and Lempicki, R. A. (2009) Bioinformatics enrichment tools: paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 37, 1−13. (17) Su, G., Kuchinsky, A., Morris, J. H., States, D. J., and Meng, F. (2010) GLay: community structure analysis of biological networks. Bioinformatics 26, 3135−3137. (18) Pavlos, R., Mallal, S., and Phillips, E. (2012) HLA and pharmacogenetics of drug hypersensitivity. Pharmacogenomics 13, 1285−1306. (19) Chen, P., Lin, J. J., Lu, C. S., Ong, C. T., Hsieh, P. F., Yang, C. C., Tai, C. T., Wu, S. L., Lu, C. H., Hsu, Y. C., Yu, H. Y., Ro, L. S., Lu, C. T., Chu, C. C., Tsai, J. J., Su, Y. H., Lan, S. H., Sung, S. F., Lin, S. Y., Chuang, H. P., Huang, L. C., Chen, Y. J., Tsai, P. J., Liao, H. T., Lin, Y. H., Chen, C. H., Chung, W. H., Hung, S. I., Wu, J. Y., Chang, C. F., Chen, L., Chen, Y. T., and Shen, C. Y. (2011) Carbamazepine-induced toxic effects and HLA-B*1502 screening in Taiwan. N. Engl. J. Med. 364, 1126−1133. (20) Aggarwal, R., Sharma, M., Modi, M., Garg, V. K., and Salaria, M. (2014) HLA-B *1502 is associated with carbamazepine induced Stevens-Johnson syndrome in North Indian population. Hum. Immunol. 75, 120−2. (21) Pichler, W. J., Beeler, A., Keller, M., Lerch, M., Posadas, S., Schmid, D., Spanou, Z., Zawodniak, A., and Gerber, B. (2006) Pharmacological interaction of drugs with immune receptors: the p-i concept. Allergol. Int. 55, 17−25. (22) Pichler, W. J., Naisbitt, D. J., and Park, B. K. (2011) Immune pathomechanism of drug hypersensitivity reactions. J. Allergy Clin. Immunol. 127, S74−81. (23) Shi, Y. W., Min, F. L., Qin, B., Zou, X., Liu, X. R., Gao, M. M., Wang, Q., Zhou, J. Q., and Liao, W. P. (2012) Association between HLA and Stevens−Johnson syndrome induced by carbamazepine in Southern Han Chinese: genetic markers besides B*1502? Basic Clin. Pharmacol. Toxicol. 111, 58−64. (24) Paquet, P., Jacob, E., Quatresooz, P., Jacquemin, D., and Pierard, G. E. (2007) Delayed reepithelialization and scarring deregulation following drug-induced toxic epidermal necrolysis. Burns 33, 100−104. (25) Liu, J., Wang, Y., Pan, Q., Su, Y., Zhang, Z., Han, J., Zhu, X., Tang, C., and Hu, D. (2012) Wnt/beta-catenin pathway forms a

negative feedback loop during TGF-beta1 induced human normal skin fibroblast-to-myofibroblast transition. J. Dermatol. Sci. 65, 38−49. (26) Ko, T. M., Chung, W. H., Wei, C. Y., Shih, H. Y., Chen, J. K., Lin, C. H., Chen, Y. T., and Hung, S. I. (2011) Shared and restricted T-cell receptor use is crucial for carbamazepine-induced Stevens− Johnson syndrome. J. Allergy Clin. Immunol. 128, 1266−1276 e1211. (27) McCormack, M., Alfirevic, A., Bourgeois, S., Farrell, J. J., Kasperaviciute, D., Carrington, M., Sills, G. J., Marson, T., Jia, X., de Bakker, P. I., Chinthapalli, K., Molokhia, M., Johnson, M. R., O’Connor, G. D., Chaila, E., Alhusaini, S., Shianna, K. V., Radtke, R. A., Heinzen, E. L., Walley, N., Pandolfo, M., Pichler, W., Park, B. K., Depondt, C., Sisodiya, S. M., Goldstein, D. B., Deloukas, P., Delanty, N., Cavalleri, G. L., and Pirmohamed, M. (2011) HLA-A*3101 and carbamazepine-induced hypersensitivity reactions in Europeans. N. Engl. J. Med. 364, 1134−1143. (28) Niihara, H., Kakamu, T., Fujita, Y., Kaneko, S., and Morita, E. (2012) HLA-A31 strongly associates with carbamazepine-induced adverse drug reactions but not with carbamazepine-induced lymphocyte proliferation in a Japanese population. J. Dermatol. 39, 594−601. (29) Yang, C. W., Hung, S. I., Juo, C. G., Lin, Y. P., Fang, W. H., Lu, I. H., Chen, S. T., and Chen, Y. T. (2007) HLA-B*1502-bound peptides: implications for the pathogenesis of carbamazepine-induced Stevens−Johnson syndrome. J. Allergy Clin. Immunol. 120, 870−877. (30) Wei, C. Y., Chung, W. H., Huang, H. W., Chen, Y. T., and Hung, S. I. (2012) Direct interaction between HLA-B and carbamazepine activates T cells in patients with Stevens−Johnson syndrome. J. Allergy Clin. Immunol. 129, 1562−1569 e1565.

H

DOI: 10.1021/tx5005248 Chem. Res. Toxicol. XXXX, XXX, XXX−XXX