Dapsone‐ and nitroso dapsone‐specific activation of T cells from hypersensitive patients expressing the risk allele HLA‐B*13:01

Abstract Background Research into drug hypersensitivity associated with the expression of specific HLA alleles has focussed on the interaction between parent drug and the HLA with no attention given to reactive metabolites. For this reason, we have studied HLA‐B*13:01‐linked dapsone hypersensitivity to (a) explore whether the parent drug and/or nitroso metabolite activate T cells and (b) determine whether HLA‐B*13:01 is involved in the response. Methods Peripheral blood mononuclear cells (PBMC) from six patients were cultured with dapsone and nitroso dapsone, and proliferative responses and IFN‐γ release were measured. Dapsone‐ and nitroso dapsone‐specific T‐cell clones were generated and phenotype, function, HLA allele restriction, and cross‐reactivity assessed. Dapsone intermediates were characterized by mass spectrometry. Results Peripheral blood mononuclear cells from six patients and cloned T cells proliferated and secreted Th1/2/22 cytokines when stimulated with dapsone (clones: n = 395; 80% CD4+ CXCR3hiCCR4hi, 20% CD8+CXCR3hiCCR4hiCCR6hiCCR9hiCCR10hi) and nitroso dapsone (clones: n = 399; 78% CD4+, 22% CD8+ with same chemokine receptor profile). CD4+ and CD8+ clones were HLA class II and class I restricted, respectively, and displayed three patterns of reactivity: compound specific, weakly cross‐reactive, and strongly cross‐reactive. Nitroso dapsone formed dimers in culture and was reduced to dapsone, providing a rationale for the cross‐reactivity. T‐cell responses to nitroso dapsone were dependent on the formation of a cysteine‐modified protein adduct, while dapsone interacted in a labile manner with antigen‐presenting cells. CD8+ clones displayed an HLA‐B*13:01‐restricted pattern of activation. Conclusion These studies describe the phenotype and function of dapsone‐ and nitroso dapsone‐responsive CD4+ and CD8+ T cells from hypersensitive patients. Discovery of HLA‐B*13:01‐restricted CD8+ T‐cell responses indicates that drugs and their reactive metabolites participate in HLA allele‐linked forms of hypersensitivity.


| INTRODUC TI ON
Genome-wide association studies have identified strong associations between the expression of HLA alleles and susceptibility to drug hypersensitivity. 1 These data suggest that drugs bind selectively to the HLA protein to activate the T cells that participate in the adverse event.
Molecular docking studies support this concept 2,3 ; however, modeling data have to be interpreted with caution as the nature of the drug-HLA protein interaction and the requirement for a specific peptide in the binding groove has not been determined. The most robust genetic associations are between HLA class I alleles and abacavir hypersensitivity, 4 flucloxacillin liver injury (both HLA-B*57:01), 5 carbamazepine-induced Stevens Johnson syndrome (HLA-B*15:02), 6 and allopurinol hypersensitivity (HLA-B*58:01), 7 and in each case, mechanistic studies have shown that CD8 + T cells are activated when the drug interacts with the relevant HLA protein. 7,8 For abacavir and carbamazepine, the drug HLA binding site is very different; abacavir binds deep in the HLA peptide binding pocket, while carbamazepine binds to a site closer to the T-cell receptor interface. Despite this, both drugs interact with HLA proteins via a reversible interaction to stimulate T cells. T cells from patients with allopurinol hypersensitivity are activated with a stable metabolite, oxypurinol, also via a direct binding interaction with HLA. In contrast, flucloxacillin-specific T cells from patients with liver injury are activated with drug-protein adducts, via a hapten mechanism involving the spontaneous binding of the drug to protein and antigen processing. This brief discussion illustrates that rapid progress has been made in our understanding of the relationship between drug HLA binding and the activation of T cells; however, reactive drug metabolites have been ignored in the study of HLA allele-restricted forms of drug hypersensitivity. This is primarily because of the absence of synthetic reactive metabolites for functional studies with T cells from hypersensitive patients.
A reactive metabolite of sulfamethoxazole is known to activate patient T cells via a hapten mechanism 12,13 ; however, sulfamethoxazole reactions are not associated with a specific HLA allele. Thus, we recently synthesized the nitroso metabolite of dapsone and studied the priming of naïve T cells from healthy donors. 15 Dapsone contains a sulfone group that links two aromatic amine moieties. Oxidative metabolism of the amine groups generates a hydroxylamine. The hydroxylamine undergoes spontaneous oxidation to form nitroso dapsone, which binds covalently to cellular proteins. 16,17 Naïve CD4 + and CD8 + T cells from healthy donors are activated with the parent drug and nitroso metabolite when Tregs were removed and the compounds were presented by dendritic cells. 15 These data show that both forms of the drug interact with multiple HLA molecules and have the capacity to stimulate T cells when regulatory pathways have been manipulated.
Dapsone is used in combination with other drugs for the treatment of infectious diseases such as leprosy and malaria. 0.5%-3.6% of treated patients develop a hypersensitivity syndrome characterized by fever, skin rash, and internal organ involvement 4-6 weeks after treatment commences. 18 HLA-B*13:01 is associated with the development of dapsone hypersensitivity in Chinese and Thai patients, 19,20 and modeling data suggest that dapsone may fit in the peptide recognition site of HLA-B*13:01. 21 Dapsone-treated cell lines expressing HLA-B*13:01 have been shown to activate T cells, while peripheral blood mononuclear cells (PBMC) from 2/7 patients secrete high levels of the cytolytic molecule granulysin when stimulated with the drug. 22 Despite this, a detailed analysis of the phenotype and function of dapsone-specific T cells has not been performed. Moreover, the activation of patient T cells with nitroso dapsone has not been investigated. Thus, our study had three primary objectives: to investigate whether dapsone and/or nitroso dapsone activates CD4 + and CD8 + T cells from hypersensitive patients, to define phenotype and function of drug-specific T cells, and to explore whether HLA-B*13:01 is directly involved in the drug-/drug metabolite-specific T-cell response.

| Human subjects
Venous blood (50 mL) was collected from 6 dapsone hypersensitive patients. Table 1 summarizes the demographics of the patients. Table   S1 shows results of HLA typing. HLA-B*13:01+ donors (n = 4) with no history of dapsone exposure and HLA-B*13:01+ dapsone-tolerant patients (n = 4) were selected as a control groups. Patch testing was conducted on the back of patients with dapsone (0.1%-25%).
The patch was removed after 48 hours. Results were recorded after a further 24 hours to exclude any false-positive responses resulting from the patch tape. The study was approved by the Ethical Committee of the Shandong Provincial Institute of Dermatology and Venereology, and informed written consent was obtained. A material transfer agreement was signed prior to shipment of PBMC to Liverpool.

| Statistics
All statistical analysis (one-way ANOVA unless stated otherwise) was performed using SigmaPlot 12 software (*P < 0.05).

| Patch testing and in vitro activation of hypersensitive patient PBMC
After 72-hour drug exposure, 3 out of 6 patients displayed dapsone concentration-dependent positive readings ( Figure 1A). PBMC from all 3 patch test-positive patients were stimulated to proliferate strongly in the presence of dapsone and nitroso

G R A P H I C A L A B S T R A C T
Dapsone and nitroso dapsone activate CD4 + and CD8 + T cells from hypersensitive patients via different mechanisms. Dapsone and nitroso dapsone interact with HLA-B*13:01 to activate certain CD8 + clones. Dapsone hypersensitivity should be used as an example to solve the structure of drug metabolite-modified peptide-HLA interactions and the relationship between HLA binding and T-cell activation. dapsone ( Figure 1B). Positive dapsone-specific proliferative responses (stimulation index [SI] 2 or above) and/or IFN-γ secretion were also detected with PBMC from the 3 patch test-negative patients, while nitroso dapsone responses were detected in 2 patients ( Figure 1B-D). Of note, patient 7 displayed a negative nitroso dapsone lymphocyte transformation test response on initial testing ( Figure 1B) and a weak response when the assay was repeated ( Figure 1D). PBMC were not activated with the co-medications rifampicin and clofazimine ( Figure 1D). PBMC from dapsone-naïve (SI < 2) and dapsone-tolerant ( Figure S1) HLA-B*13:01+ controls proliferated in the presence of phytohemagglutinin, but not the test drugs.
One hundred and two well-growing clones were selected for dose-titration studies and the analysis of cytokine secretion.

| Dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones display three distinct patterns of cross-reactivity
Sixty-three dapsone-and 98 nitroso dapsone-responsive CD4 + and CD8 + clones were assayed for cross-reactivity. When all of the dapsone-responsive clones were assessed together, low levels of proliferation were observed with nitroso dapsone ( Figure 3A).
The maximum concentration of nitroso dapsone used was 10 times lower than the dapsone concentration. Analysis of individual clones revealed three distinct cross-reactivity patterns: dapsone-specific, and weakly and strongly cross-reactive with nitroso dapsone. F I G U R E 1 Diagnosis of dapsone hypersensitivity by skin testing and in vitro assays. A, Skin patch test results for dapsone hypersensitivity. Skin was exposed to dapsone in polyethylene glycol 200 at dilutions of 0%, 0.1%, 0.5%, 1%, 5%, 10%, 15%, 2%, and 25%. The patch tape was left to the skin for 48 h, and diagnosis was made 24 h later. Red ovals show areas of inflamed skin. B, Peripheral blood mononuclear cells (PBMC) from patients were exposed to graded concentrations of dapsone or nitroso dapsone. Proliferation was measured after 5 d by the addition of [ 3 H]thymidine for 16 h. Results are expressed as mean ± SD cpm of triplicate cultures. A doubling of cpm in drugtreated cultures over vehicle control is considered positive. C, PBMC from patients were exposed to optimal concentrations of dapsone or nitroso dapsone, and IFN-γ release was visualized by ELIspot. D, PBMC from patients were exposed to graded concentrations of dapsone, nitroso dapsone, rifampicin, and clofazimine. Proliferation was measured as described in (B) (DDS, dapsone; DDS-NO, nitroso dapsone; PHA, phytohaemagglutinin) Approximately 90% of the clones were dapsone-specific or weakly cross-reactive ( Figure 3B,C).
Nitroso dapsone CD4 + and CD8 + clones displayed a much higher level of cross-reactivity ( Figure 3A). However, 3 patterns of crossreactivity were again observed with individual clones: nitroso dapsone-specific, and weakly and strongly cross-reactive with dapsone.
In contrast to the dapsone-responsive clones, approximately 90% of the nitroso dapsone-responsive clones were nitroso dapsone-specific or highly cross-reactive ( Figure 3B,C).
Dapsone-and nitroso dapsone-responsive clones were also stimulated to proliferate with dapsone hydroxylamine; however, proliferative responses were not detected when clones were cultured with (a) dapsone analogues with substitutions in the sulfone group, (b) dapsone analogues with amine groups in different positions on the aromatic rings, and (c) structurally distinct sulfonamide antimicrobials ( Figure S3).
The stability of nitroso dapsone in the proliferation assay was assessed. Nitroso dapsone was converted rapidly to azoxy dimers and the parent compound. Both compounds were detectable within 10 minutes. After 2 days, 50% of nitroso dapsone had been converted to dapsone ( Figure 3D-E).

| Cross-reactive clones are activated with equivalent concentrations of dapsone and nitroso dapsone
Six strong and weakly cross-reactive clones were incubated with dapsone and nitroso dapsone at concentrations of 0.1-100 µmol/L to define the minimum stimulatory concentrations of the two compounds. Clones were stimulated to proliferate with equivalent concentrations of each compound irrespective of the extent of cross-reactivity ( Figure 4).

| HLA-restricted activation of dapsone-and nitroso dapsone-responsive clones ensues via different mechanisms
Stimulation of dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones was dependent on the presence of antigen-presenting cells ( Figure 5A). Use of blocking antibodies revealed that CD4 + and CD8 + proliferative responses to dapsone and nitroso dapsone were HLA class II and I restricted, respectively ( Figure 5B). Fixation of antigen-presenting cells with glutaraldehyde had no effect on the activation of CD4 + or CD8 + clones with dapsone. In contrast, antigenpresenting cell fixation reduced the extent of proliferation with the nitroso metabolite ( Figure 6A). The residual response detected with nitroso dapsone and fixed antigen-presenting cells relates to the conversion of nitroso dapsone to dapsone in culture.
Nitroso dapsone-responsive CD8 + clones were activated with antigen-presenting cells pulsed with nitroso dapsone for 0.5-2 hours ( Figure 6B). Two out of 4 nitroso dapsone-responsive CD4 + clones were also stimulated to proliferate with nitroso dapsone-pulsed Phenotype and drug specificity    Table   S2. The B-cell lines were used to explore the requirement for HLA-B*13:01 in dapsone-and nitroso dapsone-specific CD8 + T-cell activation. Autologous antigen-presenting cells and antigen presenting from 1 additional patients were used as comparators.
As has been described previously for other drugs, 25,26 30% of dapsone and nitroso dapsone-responsive HLA class I restricted CD8 clones displayed proliferative responses with the drug or metabolite F I G U R E 3 Cross-reactivity of dapsone-and nitroso dapsone-responsive T-cell clones. Dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones were cultured with irradiated autologous EBV-transformed B cells and either dapsone (100-500 µmol/L) or the nitroso metabolite (5-20 µmol/L) in triplicate cultures for 48 h. T-cell proliferative responses were assessed through the addition of [ 3 H]thymidine. A, Mean cross-reactivity data for 153 clones divided according to the drug antigen peripheral blood mononuclear cells were cultured with to generate clones and CD phenotype. B, Pie charts showing number of clones with a particular cross-reactivity profile (compound specific, weekly cross-reactive (cross-reactive compound displaying 10%-50% response detected with comparator) and strongly cross-reactive (cross-reactive compound displaying > 50% response detected with comparator). C, Representative clones displaying each response profile. D, Relative quantification of dapsone and azoxy dimer in cultures containing nitroso dapsone ( Figure 7C). F I G U R E 5 Antigen-presenting cells are required for the activation of dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones. A, Dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones were cultured with dapsone (500 µmol/L) or the nitroso metabolite (20 µmol/L) in triplicate cultures for 48 h either in the presence or absence of autologous EBV-transformed B cells. B, Dapsone-and nitroso dapsone-responsive CD4 + and CD8 + clones were cultured with antigen-presenting cells and dapsone (500 µmol/L) or the nitroso metabolite (20 µmol/L) in triplicate cultures for 48 h either in the presence or absence of anti-HLA class I and II blocking antibodies.

| D ISCUSS I ON
Knowledge of the role drug metabolism plays in the generation of antigenic determinants that activate T cells is limited. Circumstantial evidence supporting a role for drug metabolism includes the identification of protein-reactive metabolites for many drugs associated with a high incidence of hypersensitivity and the induction of toxicity in target tissue by reactive metabolites and hence the potential to disrupt immune regulatory pathways through the provision of danger signals. 27 However, the most direct evidence linking drug metabolism to the development of hypersensitivity is the detection of halothane-specific antibodies in patients with liver failure 28  In recent years, researchers studying hypersensitivity reactions strongly linked to expression of specific HLA alleles have focussed exclusively on the interaction between parent drug and the HLA molecule. This is because the parent drug is, for the most part, the   for T-cell activation.
25%-50% of CD4 + and CD8 + clones were classified as highly specific as they were only activated with one compound (either the parent drug or metabolite [ Figure 3B]). This confirms that T cells recognize and respond selectively to the two different forms of the dapsone antigen. Dapsone has an unusual structure in that the two aromatic amines connected to the sulfone group are identical.
If the structure of a nitroso dapsone-modified HLA binding peptide is compared with dapsone complexed to the same peptide, it is likely that similar conformations will be observed, and as such, the interaction with the T-cell receptor will be the same. Thus, our working hypothesis to explain the drug-or drug metabolite-specific activation of certain clones is that the HLA binding peptides also participate in the TCR interaction and impart a degree of selectivity.
When using optimum concentrations of dapsone (100-500 µmol/L) and nitroso dapsone (5-20 µmol/L) to activate CD4 + and CD8 + T cells, clones displaying high (ie, at least 50% of the response detected with the opposite compound) and low (ie, 10%-50% of the response detected with the opposite compound) levels of cross-reactivity were also detected. The majority of dapsone-responsive, cross-reactive CD4 + and CD8 + clones displayed low levels of cross-reactivity with nitroso dapsone. Using quantitative mass spectrometry, we were able to demonstrate that the nitroso metabolite is reduced to dapsone in the 2-day proliferation assay. Thus, clones incubated with 30 µmol/L nitroso dapsone were were activated with nitroso dapsone-pulsed antigen-presenting cells and the strength of the induced response was the same as that seen with the parent drug. Two of the nitroso dapsone-responsive clones were not activated with pulsed antigen-presenting cells; interestingly, both of these clones cross-reacted strongly with dapsone.
Glutathione is a tripeptide intracellular antioxidant that protects cells from exposure to aromatic nitroso compounds by acting as a reducing agent and through direct conjugation. 23,24 The addition of glutathione to T-cell assays prevents the covalent binding of nitroso compounds, and hence, it is possible to explore whether T cells are activated with drug metabolite-modified protein adducts. Thus, ELIspot was used to profile cytokine secretion from dapsoneand nitroso dapsone-stimulated clones. CD4 + and CD8 + clones secreted the same panel of cytokines when stimulated with dapsone or nitroso dapsone, but expressed distinct chemokine receptors.
Clones secreted Th1, Th2, and Th22 cytokines alongside the cytolytic molecules perforin, granzyme B, and FasL; however, IL-17 was not detected. IL-22 secretion in the absence of IL-17 seems to be a common feature of drug-specific clones as this profile has now been detected with dapsone, sulfamethoxazole, and piperacillin. 42,43 It is possible that this profile relates to the long-term in vitro culture of clones; thus, a future study exploring the cytokine profile in affected tissues would be appropriate. The chemokine receptors displayed on dapsone-or nitroso dapsone-responsive CD4 + clones were restricted to CXCR3 and CCR4. In contrast, CD8 + clones also expressed CCR6, 9, and 10. CCR4 and CCR10 have been implicated in the migration of T cells into skin 44 ; thus, a more detailed investigation of migratory properties of CD4 + and CD8 + T cells in patients with dapsone hypersensitivity is warranted.
In conclusion, our study shows that dapsone-and nitroso dapsone-responsive CD4 + and CD8 + T cells circulate in hypersensitive patients. T cells were activated selectively with the parent drug and drug metabolite via direct HLA binding and a hapten mechanism, respectively. The detection of HLA-B*13:01-restricted dapsone-and nitroso dapsone-responsive CD8 + clones indicates that dapsone hypersensitivity should be used as an exemplar to explore the structural features of drug HLA binding and how this interaction results in a pathogenic T-cell response. For patients with suspected hypersensitivity, it is critical to identify the agent that caused the reaction.
We are currently exploring whether drug-peptide antigens represent effective reagents for T-cell activation. A better understanding of drug antigenicity will lead to the development of improved evidence-based diagnostic tools.

ACK N OWLED G M ENTS
The authors would like to thank the patients and volunteers for agreeing to donate blood and tissue samples.

CO N FLI C T S O F I NTE R E S T
The authors declare that they have no conflicts of interest.