Study design and patient recruitment

From 2012 to 2020, we collected twelve clinical information, as well as eleven blood cells samples of DRESS and one blister cells sample of SJS/TEN from patients diagnosed with DDS-induced hypersensitivity reactions (DDS-DIHS), at the Institute of Dermatology, Chinese Academy of Medical Sciences, Nanjing and the Chang Gung Memorial Hospitals, Taiwan. DRESS was diagnosed according to the RegiSCAR criteria [34, 35] and was characterized by extensive erythematous maculopapular rash with periorbital edema and desquamation. The diagnosis of DRESS using the criteria and scoring system of the RegiSCAR group include cutaneous involvement with typical skin eruptions (e.g., exfoliative dermatitis, generalized maculopapular exanthema), fever (≧ 38.5 °C), enlarged lymph nodes (≧ 2 sites, ≧ 1 cm), presence of atypical lymphocytes and eosinophilia, systemic involvement (e.g., liver, kidney, and lung), time of resolution, and the evaluation of other potential causes. The clinical course, dosage and duration of dapsone, systemic involvement, and mortality were analyzed. The drug causality of SJS and DIHS was determined by the algorithm of drug causality for epidermal necrolysis (ALDEN) and Naranjo algorithm following the guidelines for assessment of drug causality published by the RegiSCAR group [35,36,37,38]. Only the cases with probable or certain causal association with DDS (ALDEN score greater than 4 or Naranjo algorithm greater than 5 were recruited. All DDS-DIHS patients and one DDS-SJS patient were HLA-B*13:01-positive, and were assessed by two dermatologists through review of photographs, pathology tissue slides, and medical records. We also recruited ten DDS-tolerant control subjects who had been administered DDS for at least 3 months without any cutaneous adverse reactions. Among them, three were HLA-B*13:01-positive, and seven were HLA-B*13:01-negative leprosy patient (see Table 1). In addition, twelve healthy control individuals were enrolled. Approvals were obtained from the institutional review board (IRB No. 100-4657A3, 201601761B0, 201902171A3 and 202001645B0), and informed consent was obtained from each participant.

Table 1 Clinical characteristics of twelve patients and ten tolerant controls with dapsone-induced hypersensitivity reactionsChemicals, vectors and cell lines

DDS, the DDS metabolite N-acetyl DDS (NAD), sulfadiazine and trichloroethylene were purchased from Sigma-Aldrich (St Louis, Mo). EBV-transformed autologous B-cell lines (B-LCLs), and HLA class I-deficient lymphoblastoid B-cell line (Hmy2.C1R), which were used as antigen-presenting cells (APC), were obtained from ATCC (Maryland, USA). For mutation assays, full-length open-reading frames encoding HLA-B*13:01, HLA-B*13:02 and β2M cloned into VP64-GFP vector. The vectors encoding wild-type HLA-B*13:01 and HLA-B*13:02 were then used as templates for generation of single-site and two-site HLA-B*13:01 and HLA-B*13:02 mutant variants with site-directed mutagenesis (Table 2). The VP64-GFP vectors encoding wild-type or one of mutant HLA-B*13:01 or HLA-B*13:02 variants were then transfected along with VP64-GFP-β2M, pspax2 and pMD2.G to Hmy2. C1R cells (HLA-B-deficient cells) and sorted by flow cytometry with GFP fluorescent protein as a reporter. Stable expression of the C1R transmembrane HLA-B*13:01 (C1R-HLA-B*13:01) clone was determined using flow cytometry by determining protein expression with anti-HLA antibody (w6/32, eBioscience).

Table 2 X-ray crystallography data collection and refinement statisticsExpression and purification of recombinant proteins for crystallographic experiments

Gene sequences for HLA-B*13:01 and beta-2-microglobulin (GenBank ID: AAA59627.1 and AAA51811.1) were synthesized at Sangon Biotech (China). The genes were cloned into the pET21a vector with an C-terminal 6 × His tag and expressed in Escherichia coli BL21 (DE3) cells. The cells were grown in LB medium at 37 °C until the OD600 nm reached 0.6, after which the expression of the recombinant proteins was induced at 16 °C for 20 h with the final concentration of isopropyl-B-D-1-thiogalactopyranoside (IPTG) of 0.1 mM. The cells were then harvested by centrifugation at 12,000 × g for 20 min, resuspended in lysis buffer containing 6 M guanidine-HCl and ultrasonicated. Recombinant proteins (180 mg HLA-B*13:01 and 60 mg β2M) were renatured overnight at 4 °C in dialysis buffer (0.1 M Tris pH = 8.0, 2 mM EDTA, 400 mM l-arginine-HCl, 0.5 mM oxidized glutathione), following which the precipitate was removed by centrifugation. The clear supernatant was loaded onto a 5 mL nickel-nitrilotriacetic acid (Ni–NTA) resin gravity column (Qiagen) pre-equilibrated with binding buffer (20 mM Tris–HCl pH = 8.0, 150 mM NaCl). The column was successively washed with 50 mL binding buffer containing 20 mM imidazole and 100 mM imidazole, and target protein was eluted using binding buffer containing 500 mM imidazole. The target proteins were then concentrated and loaded onto a Superdex G200 size-exclusion chromatography (SEC) column (120 mL, GE Healthcare Life Sciences, USA) pre-equilibrated with SEC buffer (20 mM Tris–HCl pH 8.0, 150 mM NaCl and 2 mM DTT). As initial SEC results indicated that the majority of HLA-B*13:01 did not bind to β2M, we used different oligopeptides predicted by NetMHC 4.0 Server (http://www.cbs.dtu.dk/services/NetMHC/) to stabilize the HLA-B*13:01-β2M complex. After purification of HLA-B*13:01 and β2M, the predicted peptides were added respectively. The formation of HLA-B*13:01-β2M complex was verified by SEC combined with SDS-PAGE, and finally found that the peptide (RQDILDLWI) could promote the formation of the complex.

Crystallization

Initial crystallization screens for the HLA-B*13:01-β2M-peptide complex (concentrated to 5–7 µg/mL) were performed with commercial crystallization screening kits using the sitting-drop vapor diffusion method at 16 °C. Crystallization drops contained 0.5 μL of the protein complex solution mixed with 0.5 μL of reservoir solution. Diffraction quality crystals were grown in 0.1 M HEPES (pH 6.5–7.5), 20% (v/v) PEG 4000, and 0.2 M NaCl. All crystals were cryoprotected in the reservoir solution supplemented with 20–25% (v/v) glycerol and flash-frozen in liquid nitrogen.

Data collection and structure determination

X-ray diffraction data for the HLA-B*13:01-β2M-peptide complex crystal was collected at the beamline BL-17U1 of the Shanghai Synchrotron Radiation Facility (SSRF). All data were indexed and scaled using HKL2000 software. The structure of HLA-B*13:01-β2M-peptide complex was determined by molecular replacement with MOLREP in PHENIX [39] using the structure of HLA-B*44:03 in complex with Epstein-Barr virus BZLF1-derived peptide (PDB ID: 4JQX) as the search model. AutoBuild in PHENIX was used to automatically build the structure model. After several rounds of positional and B-factor refinement using Phenix. Refine with TLS parameters alternated by manual model revision in Coot [40], model quality was verified using the PROCHECK program (https://www.ebi.ac.uk/thornton-srv/software/PROCHECK). The quality of the final model was validated with MolProbity [41]. Structures were analyzed with the PDBePISA (Protein Interfaces, Surfaces, and Assemblies) tool on Dali server (http://ekhidna2.biocenter.helsinki.fi/dali). Detailed information on data collection and refinement statistics is listed in the Table 2. All structural figures were prepared in PyMOL (http://www.pymol.org).

Surface plasmon resonance

Biacore T200 surface plasmon resonance (SPR) system (GE Healthcare, Piscataway, NJ) was used to analyze the interaction between HLA-B*13:01 and different drugs. The recombinant HLA-B*13:01 and β2-microglobulin (Chemicon International, Temecula, Calif) were incubated together in PBS buffer and then immobilized on the chips via amine coupling reaction. The drugs used for the analysis were dissolved in PBS supplemented with 5% dimethyl sulfoxide (DMSO and flowed through the solid phase. Responses of the interaction were reference subtracted and corrected with a standard curve for the DMSO effects. BIA Evaluation software Version 1.0 was used for data analysis. The Kd value were determined by three independent experiments.

HLA-B*13:01 fix and pulse and binding assay

We investigated the DDS/HLA-B*13:01 interaction based on cytotoxic T lymphocytes (CTLs) activity. Specific CTLs reacted even if the APCs were fixed by 0.25% paraformaldehyde, excluding that either processing or intracellular metabolism is involved. Pulsed C1R-HLA-B*13:01 were incubated with the DDS for 1 h followed by two washing steps, no stimulation of drug-specific T-cell clones were observed for DDS. A C1R-HLA-B*13:01 stable clone was generated by means of transfection of the full-length cDNA plasmid encoding HLA-B*13:01. The endogenous peptides in HLA-B*13:01 were removed by using cold mild citric acid. After neutralizing with culture medium, the cells were incubated with β2M (4 µg/mL), GolgiStop (1 µg/mL; BD Biosciences, Calif), and the synthetic peptides for 3 h at room temperature.

HLA-B*13:01 protein extraction, co-immunoprecipitation (Co-IP) and peptides identification by liquid chromatograph-mass spectrometer (LC–MS)

C1R and C1R-HLA-B*13:01 were cultured with DDS (50 µg/mL) for 6 h and collected cell pellet to add RIPA lysis buffer. The supernatants were collected by centrifuge at 10,000 rcf for 10 min and transferred to a 3 kD ultrafiltration tube and centrifuge at 12,000 rcf for 10 min at 4 °C, then collected less than 3 kD peptides. The protein samples were testing by HLA-ABC antibody immobilization and Co-IP assay and analyzed by Easy-nLC1000 and Q Exactive™ Hybrid Quadrupole-Orbitrap™ Mass Spectrometer (Thermo Fisher Scientific, USA). The raw MS files were analyzed and searched against protein database based on the species of the samples using PEAKS Studio. The parameters were set as follows: the protein modifications were carbamidomethylation (C) (fixed), oxidation (M) (variable), Acetyl (Protein N-term) (variable); the enzyme specificity was set to Unspecific; the maximum missed cleavages were set to 2; the precursor ion mass tolerance was set to 20 ppm, and MS/MS tolerance was 20 ppm. Only high confident identified peptides were chosen for downstream protein identification analysis.

Growth of DDS-specific T cells

Peripheral blood mononuclear cells (PBMCs) were isolated from whole blood samples using Ficoll-Paque (GE, Life Sciences, USA) density gradient centrifugation. The PBMCs (5 × 10e5 per well) of patients with DDS-induced SCAR and other control individuals were cultured in 96-well microplates in TexMACS medium (miltenyi Biotec, Germany) supplemented with 5% human AB serum [42] (Sigma-Aldrich, Darmastadt, Germany), penicillin–streptomycin (Gibco Invitrogen, USA), 200U IL-2 (R&D systems, USA), and 50 µg/mL DDS (Sigma-Aldrich, St. Louis, MO) for 1 week at 37 °C in 5% CO2. DDS specific T-cell lines (TCLs) were obtained by culturing the patients’ PBMCs with DDS (50 µg/mL) for 12–14 days, and the expanded T cells were then restimulated with irradiated (50 Gy) autologous B-LCLs and DDS for approximately 4 to 5 cycles. The T-cell clones were obtained by means of serial dilution. Ten CTL TCLs were sorted by using FACSAria (BD, Franklin Lakes, NJ). In addition, dimethyl sulfoxide (DMSO) was added to the medium as the solvent control, and phytohemagglutinin (i.e., PHA) at a concentration of 10 µg/mL was used as the positive control. Proliferative clonotypes of T cells were performed to obtain for T-cell receptors (TCR) study.

TCR repertoire analysis by high-throughput next generation sequencing and single-cell sequencing

Proliferative clonotypes of T cells from DDS-induced SCAR patients were performed for (1) normal TCR sequencing based approach, that RNA was purified from each clone using Qiagen RNeasy, and 5′RACE was performed using BD SmartRace reagents and protocol, using the universal 5′primer, and a 3′gene-specific primer for the TCR constant region. (2) For experiments using the 10 × Genomics platform, Human T Cell (PN-1000005) were used according to the manufacturer’s instructions in the Single Cell V(D)J Reagent Kits User Guide. Cell number and concentration was confirmed with TC20™ Automated Cell Counter. Approximately Number (5000, etc.) cells were subjected immediately onto the 10XGenomics Chromium Controller machine for Gel Beads-in-Emulsion (GEM) generation. Full-length V(D)J segments from either T cell transcripts are enriched from first-strand cDNA via PCR amplification with primers specific to either the TCR or Ig constant regions prior to library construction. Library quality and concentration were assessed using Agilent Bioanalyzer 2100. Libraries were run on the Hiseq X or Novaseq for Illumina PE150 sequencing. Post-processing and quality control were performed by Novogene using the 10X Cell Ranger package (v2.2.0, 10X Genomics). Reads were aligned to mm10 (or GRCh38, etc.) reference assembly (v2.2.0, 10X Genomics).

Generation of single chain TCRα/TCRβ expression constructs and cells

Retroviruses encoding TCR genes carrying human TCR constant regions had the format TCRα-F2A-TCRβ and were produced in HEK-293T cells by transient transfection of retroviral-based plasmids and their packaging vectors (psPAX2 and pMD2.G) using lipofectamine 3000 (Thermo, Life, USA) according to the manufacturer’s protocol. At 48 h after transfection, the virus was collected, filtered through a 0.45-µm syringe filter, and used for infection. The T cells from PBMC were spin-infected with viral supernatant supplemented with 10 µg/mL Polybrene at 2500 r.p.m. at 30 °C for 90 min. On day 3 post-infection, TCR expressing cells were sorted by flow cytometry to establish derivative cell lines as indicated.

Flow cytometry

Flow cytometry was carried out using distinct fluorochrome-conjugated mAbs that recognize human CD3, CD4, and CD8 (BD Biosciences), human TCRα, TCRβ (BioLegend). These mAbs were labeled with Alexa Fluor 488, phycoerythrin (PE), phycoerythrin–cyanine 5 (PC5), phycoerythrin-cyanin 7 (PC7) or allophycocyanine. The cells were examined by means of multicolor flow cytometry on the BD FACSVerse flow cytometer (BD Biosciences), and data were analyzed with the Flowjo™ V10 software (BD Biosciences).

T cell activation and transduction

To transduce primary human T cells, PBMC (2 × 10e6 cells/mL) were activated in 24-well plates coated with 1 µg/mL antibody to CD3 (clone OKT3, eBioscience) in the presence of 1 µg/mL soluble antibody to CD28 (clone CD28.2, eBioscience) and 300U/mL interleukin-2 (IL-2; Fisher Scientific) in TexMACS medium supplemented with 5% (v/v) human AB serum and antibiotics (penicillin–streptomycin). After 48 h of activation, the cells were spin-infected with viral supernatant supplemented with 10 µg/mL Polybrene for 90 min at 2500 rpm at 30 °C. After spin-infection, the retroviral supernatant was replaced with fresh T cell medium containing 300 U/mL IL-2 and 1 µg/mL anti-CD28. The transduced primary T cells were cultured for 48 h and then used for cytotoxicity assays.

ELISpot assay

PBMCs isolated from DDS-SCAR patients were incubated with DDS for 1 week. The DDS-specific T cells were then expanded by adding 50 µg/mL DDS, 30 nanogram/mL CD28, 30 nanogram/mL OKT3 (anti-CD3 antibody), and 200 U/mL IL-2 for 2 weeks. The expanded T cells were reset for 3 days, and CTLs were isolated using FACS with anti-human CD45, CD3, and CD8 antibodies as markers. The DDS-specific CTLs were detected by quantifying the IFN-γ release with ELISpot assay following the instructions of the human IFN-γ ELISpot kit (BD Biosciences, USA). To perform ELISpot analysis, 5 × 10e4 DDS-specific CTLs or TCR transfectants were incubated with 1 × 10e4 C1R, C1R-HLA-B*13:01 or EBV-B-LCL-13:01 APC at 37 °C for 24 h. Data was analyzed using the CTL ELISpot Reader (CTL, USA).

Real-time RT-PCR assay

Total RNA extracted from patients’ PBMCs using Qiagen RNA Extraction Kit and was reverse transcribed using the Superscript II reverse transcriptase enzyme (Life Technologies). RT-PCR was performed in 20 µL reactions containing 0.8 µM forward and reverse primers and 5 µL of extracted RNA on an ABI 7300 (Thermo Fisher) using the AgPath-ID One-Step RT-PCR kit (Thermo Fisher). Thermocycle conditions included initial denaturation at 50 °C and 95 °C (10 min each), followed by 40 cycles at 95 °C (15 s) and 60 °C (1 min). The TCR primers used in quantitative-PCR amplification were showed in Additional file 3: Table S1. To verify primer specificities, melting curve analyses and PCR product sequencing were performed. Ct value were calculated via 2-ΔΔct method after normalization to GAPDH.

Modeling TCR/peptide/HLA and DDS

The 3D structures of HLA-B*13:01 and a TCR clonotype were first constructed with homology modeling method Modeller independently [43]. And next the complex structure was assembled by referring the existing crystal structure (PDB code: 4EUP) using structure alignment program TMalign [44]. This complex was further refined by a protein side-chain repacking method CISRR [45] as well as energy minimization based on CHARMM22 force field [46]. The binding modes of DDS with the HLA-B*1301 were generated by protein–ligand blind docking tool CB-Dock [47] and superposed onto the complex structure of HLA-B*13:01-TCR. Images were generated with PyMOL (PyMOL Molecular Graphics System, Version 1.2, Schrodinger LLC, New York, NY).

Determination of inflammatory cytokine and cytotoxin production levels

To quantify IFN-γ, IL-3, IL-13, IL-5, IL-21, granulysin and granzyme B production levels, B-LCL (HLA-B*13:01) were plated in 24-well plates (10e5 cells/well) in co-culture with CTL (5 × 10e5 cells/well) and stimulated with DDS (50 µg/mL) in 1 mL complete RPMI1640 media overnight. Afterward, the supernatant concentrations of the inflammatory cytokines and cytotoxins mentioned above were quantified with ELISA (R&D, USA).

Statistical analysis

Statistical analyses were performed using SPSS for Windows, version 21.0 (IBM, Armonk, NY) and GraphPad prism 8.0 (San Diego, CA). Heatmaps were created using the built-in R heatmap function in stats package, and the Circos plots were generated using the VDJtools software (MiLaboratory). The entire experiment was repeated thrice. P-values for ratio estimates were calculated using a two-sided test. For analysis of ELISpot results, statistical comparison between two variables was performed by two-tailed Student’s t-test. Differences were considered statistically significant at P-values of less than 0.05.

Data availability

All sequencing data obtained in the study have been deposited in the NCBI sequence read archive (SRA) database with links to BioProject accession ID “PRJNA867422” (it will be uploaded after manuscript accepted). The atomic coordinates and structure factors of the HLA-B*13:01-β2M-peptide complex was deposited in the Protein Data Bank under the accession number 7ER5. All other data are available from the authors upon reasonable request.