Isoniazid, rifampicin, pyrazinamide, and ethambutol are commonly used for the treatment of tuberculosis. Drug exposure is occasionally associated with liver and/or skin injury. The aim of this study was to determine whether drug-specific T-cells are detectable in patients with adverse reactions and if so characterize the nature of the T-cell response. Peripheral blood mononuclear cells (PBMC) from 6 patients with anti-tuberculosis drug-related adverse reactions (4 liver, 2 skin) were used to detect drug-responsive T-lymphocytes. Positive lymphocyte transformation test and/or ELIspot results were observed with all 6 patients. Over 3400 T-cell clones were generated from isoniazid, rifampicin, pyrazinamide, or ethambutol-treated PBMC. CD4+ clones from all 3 patients were activated to proliferate and secrete cytotoxic mediators (granzyme B, perforin, FasL) and effector (IFN-γ, Il-13) and regulatory (Il-10) cytokines with isoniazid, but not rifampicin, pyrazinamide, or ethambutol. Il-17 was not detected, while only 1 clone secreted Il-22. Isoniazid-responsive clones were not activated with other anti-tuberculosis drugs or isonicotinic acid albumin adducts. Activation of the clones with isoniazid was MHC class II-restricted and dependent on antigen-presenting cells. Most clones were activated rapidly even in the presence of the enzyme inhibitor 1-aminobenzotriazole. However, a time-dependent pathway of activation involving auto-oxidation of isoniazid was also observed. The discovery of isoniazid-specific CD4+ T-cell clones in patients with liver and skin injury suggests that the adaptive immune system is involved in the pathogenesis of both forms of iatrogenic disease.

Approximately 9 million new cases of active tuberculosis (TB) are reported every year and TB is responsible for 3 deaths every minute world-wide (Galagan 2014). A combination of isoniazid (INH), rifampicin (RIF), pyrazinamide (PZA), and/or ethambutol (ETB) is commonly used for the treatment of TB (anti-tuberculosis drugs, ATDs). A significant clinical problem of ATD treatment is a mild elevation of liver enzymes that occasionally develops into severe liver injury. During standard ATD treatment, 2–28% of individuals develop mild liver injury (Agal et al. 2005). Identifying the culprit drug involved in the pathogenesis of ATD-induced liver injury is often difficult due to multiple medications being taken simultaneously. Furthermore, the mechanism of tissue injury has not been resolved. ATD treatment can also cause delayed-type skin reactions either in the presence or absence of liver injury.

Age, gender, nutritional status, alcohol intake, concomitant infection, and polymorphisms in drug metabolizing enzymes are known risk factors for ATD-induced adverse reactions (Ramappa and Aithal 2013). However, the delayed onset of ATD-induced liver and skin injury and the discovery of an HLA-risk allele (Sharma et al. 2002) suggest that the adaptive immune system might contribute to the iatrogenic diseases. Early studies by Warrington et al. found that INH and INH-modified albumin activated peripheral blood mononuclear cells (PBMC) isolated from certain patients with liver injury following INH monotherapy (Warrington et al. 1982). More recently, Metushi et al. (2014a) found anti-INH and anti-CYP auto antibodies in patients with INH-induced liver injury. The prevailing anti-INH antibody subclass was IgG3 (Metushi et al. 2014b), which fixes compliment via a pathway dependent on Th1 cytokine IFN-γ. These data suggest that INH protein adducts activate humoral and cellular responses in patients with INH monotherapy-induced liver injury; however, the nature of the drug-specific T-cell response has not been defined. Furthermore, drug antigen specificity in liver injury patients exposed to multiple ATDs has not been studied. Lehloenya et al. (2015) recently used skin testing and oral challenge to confirm an immune pathogenesis in patients with ATD-induced severe skin reactions. Individual patients developed reactions to INH or the structural analogue ethionamide, with little cross reactivity.

The objective of this study was to investigate whether a drug-specific T-cell response was detectable in patients with delayed-onset ATDs-related liver and skin injury. T-cell clones were subsequently isolated and characterized in terms of antigen specificity, cellular phenotype and function.

MATERIAL AND METHODS

Human subjects

Six patients with delayed-onset ATD-related liver or skin injury and 3 patients with TB on ATB drugs with no adverse reactions were recruited. Table 1 lists the clinical features of the reactions and concomitant medications at the time of the reaction. Eight healthy donors were used to investigate whether activation of T-cells is only observed in exposed patients. A total of 40–100 ml of blood was collected for PBMC isolation. Genomic DNA was extracted from 10 ml venous blood using Chemagic magnetic separation (Chemagen, Baesweiler, Germany). High-resolution sequence-based HLA typing using next generation massively parallel sequencing platforms, was performed by Histogenetics laboratory (Histogenetics, New York) at the following 6 loci: HLA-A, -B, -C, -DRB1, -DQB1, and DQA1. Table 2 shows ethnicity and HLA typing data for the 6 patients with ATD-related adverse reactions. Approval for the study was acquired from the Leeds local Research Ethics Committee and informed written consent was obtained from each donor.

Table 1

Clinical Details of the Patients and Lymphocyte Transformation Test/IFNγ-ELIspot Results.

ID Gender Age Diagnosis Medication at Time of diagnosisa ADR LTT/IFNγ-ELIspot
 
INH RIF ETB PZA INAb INA-NHS INA-HSA 
Liver injury             
 022 67 Latent disease INH/RIF Liver reaction week 4 (ALT 417). Resolved with drug discontinuation with no jaundice/hospitalization. −/ ++c −/− −/− −/− −/ ++ n/− n/− 
 031 18 Latent disease INH/RIF Skin reaction (week 2). Subsequent liver reaction week 4 with INH alone (ALT 208). No jaundice or hospital admission. +/− −/− n/nb n/n +/− n/n n/n 
 034 62 Pulmonary TB INH/RIF/ETB/PZA Liver reaction week 2 (ALT 345). Felt nauseous with poor appetite, some tenderness over liver. No jaundice or hospital admission  ++/ ++ −/− −/+ +/− −/− −/− n/− 
 037 70 Latent disease INH Liver reaction week 4 (ALT 709). Felt unwell with abdominal discomfort. Admitted to hospital; reaction settled without treatment +/− −/− −/− −/− −/− n/n n/n 
Skin injury 
 008 17 Pulmonary TB INH/RIF/ETB/PZA Skin reaction week 2 (ALT 17) +/+ −/− −/ ++ −/− n/− −/− n/− 
 043 79 Pulmonary TB INH/RIF/ETB/PZA Urticarial rash week 2 (ALT 21) −/− −/− −/+ −/− n/− n/− n/− 
ID Gender Age Diagnosis Medication at Time of diagnosisa ADR LTT/IFNγ-ELIspot
 
INH RIF ETB PZA INAb INA-NHS INA-HSA 
Liver injury             
 022 67 Latent disease INH/RIF Liver reaction week 4 (ALT 417). Resolved with drug discontinuation with no jaundice/hospitalization. −/ ++c −/− −/− −/− −/ ++ n/− n/− 
 031 18 Latent disease INH/RIF Skin reaction (week 2). Subsequent liver reaction week 4 with INH alone (ALT 208). No jaundice or hospital admission. +/− −/− n/nb n/n +/− n/n n/n 
 034 62 Pulmonary TB INH/RIF/ETB/PZA Liver reaction week 2 (ALT 345). Felt nauseous with poor appetite, some tenderness over liver. No jaundice or hospital admission  ++/ ++ −/− −/+ +/− −/− −/− n/− 
 037 70 Latent disease INH Liver reaction week 4 (ALT 709). Felt unwell with abdominal discomfort. Admitted to hospital; reaction settled without treatment +/− −/− −/− −/− −/− n/n n/n 
Skin injury 
 008 17 Pulmonary TB INH/RIF/ETB/PZA Skin reaction week 2 (ALT 17) +/+ −/− −/ ++ −/− n/− −/− n/− 
 043 79 Pulmonary TB INH/RIF/ETB/PZA Urticarial rash week 2 (ALT 21) −/− −/− −/+ −/− n/− n/− n/− 

Abbreviations: INA, isonicotinic acid; INA-HSA, INA coupling to albumin; INA-NHS, N-hydroxysuccinimide ester of isonicotinic acid.

a

Isoniazid (INH), rifampicin (RIF), pyrazinamide (PZA), and/or ethambutol (ETB).

b

n, not examined; c ++ stimulation index > 3; + 3 > stimulation index > 1.8; −1.8 > stimulation index.

Table 2

HLA Typing and Ethnicity of the Patients with ATD-Related Adverse Reactions

Patient ID Ethnicity HLA Alleles
 
  DRB1 DQB1 DQA1 
31 Caucasian 01:01 31:01 51:01 51:01 15:02 15:02 04:07 04:07 03:01 03:01 03:01 03:01 
34 Caucasian 30:01 68:01 13:02 39:01 06:02 12:03 03:01 13:01 02:01 06:03 01:03 05:01 
37 Caucasian 02:01 31:01 35:02 44:02 04:01 05:01 04:01 04:03 03:02 03:05 03:01 03:01 
43 Caucasian 24:02 31:01 52:01 56:01 04:01 12:02 04:03 15:02 03:02 06:01 01:03 03:01 
22 Indian 03:01 26:01 27:05 38:01 01:02 12:03 01:01 13:01 05:01 06:03 01:10 01:01 
Japanese 11:01 11:01 35:03 44:02 04:01 05:01 11:01 11:04 03:01 03:01 05:01 05:01 
Patient ID Ethnicity HLA Alleles
 
  DRB1 DQB1 DQA1 
31 Caucasian 01:01 31:01 51:01 51:01 15:02 15:02 04:07 04:07 03:01 03:01 03:01 03:01 
34 Caucasian 30:01 68:01 13:02 39:01 06:02 12:03 03:01 13:01 02:01 06:03 01:03 05:01 
37 Caucasian 02:01 31:01 35:02 44:02 04:01 05:01 04:01 04:03 03:02 03:05 03:01 03:01 
43 Caucasian 24:02 31:01 52:01 56:01 04:01 12:02 04:03 15:02 03:02 06:01 01:03 03:01 
22 Indian 03:01 26:01 27:05 38:01 01:02 12:03 01:01 13:01 05:01 06:03 01:10 01:01 
Japanese 11:01 11:01 35:03 44:02 04:01 05:01 11:01 11:04 03:01 03:01 05:01 05:01 

Chemicals

INH, RIF, PZA, ETB, isonicotinic acid (INA), human serum albumin (HSA) (approximately 97% pure, lyophilized), and other standard chemicals were obtained from Sigma-Aldrich (Poole, Dorset, UK). LC/MS grade solvents were purchased from Fisher (Loughborough, Leicestershire, UK).

Synthesis of N-hydroxysuccinimide ester of INA and modification of HSA by isoniazid

The activated ester of INA was synthesized by the reaction of isonicotinoyl chloride hydrochloride (Sigma) with N-hydroxysuccinimide (NHS, Sigma) to form INA-NHS (

) as previously described (Chowdhury et al. 2006). INA-NHS was coupled to HSA (0.6 mM, Sigma) by adding a 50-fold molar excess (based on 59 lysine residues/HSA) of INA-NHS (30 mM) in 250 μl of 0.1 M Na2HPO4 and 0.15 M NaCl buffer at 37 °C for 2 h. The unbound compound was removed with a 10 000 MW cut off filter. Finally, the protein pellet was reconstituted in RPMI for cell assays or buffer for MS-analyses of INH coupling to HSA [isonicotinic amide adducts (INA-HSA)].

LC-MS/MS analyses

Established mass spectrometric methods (Meng et al. 2015) were used to (1) characterize INH coupled HSA and (2) define the amino acid residues on albumin modified with INH in cell culture medium after 1–48 h. Relative quantification of modified peptides was performed by comparing the normalized intensity of MRM peaks for each of the modified residues against total ion counts across samples. Data were analyzed using Analyst software (AB Sciex).

Medium for T-cell culture and cloning

Culture medium consisted of RPMI-1640 supplemented with pooled heat-inactivated human AB serum (10%, vol/vol), HEPES (25 mM), L-glutamine (2 mM), transferrin (25 μg/ml), streptomycin (100 μg/ml), and penicillin (100 U/ml).

Detection of anti-tuberculosis drug-specific peripheral blood mononuclear cell responses

Proliferation of patient and healthy donor PBMC (1.5 × 105 cell/well) in culture medium against INH (0.01–4 mM), INA (0.01–1 mM), RIF (0.004–0.4 mM), ETB (0.01–4 mM), PZA (0.01–4 mM), INA-NHS (0.01–2 mM), INA-HSA (0.5–4 mg/ml), and tetanus toxoid (5 μg/ml, as a positive control) was measured using the lymphocyte transformation test (LTT) (Whitaker et al. 2011). Interferon-gamma (IFN-γ) secreting PBMC were visualized using ELIspot (MabTech, Nacka Strand, Sweden) by culturing cells (5 × 105 cell/well; 200 µl) in culture medium for 48 h with INH (0.1–2 mM), INA (0.1–1 mM), RIF (0.01–0.1 mM), ETB (0.1–1 mM), PZA (0.1–2 mM), INA-NHS (0.1–1 mM), INA-HSA (0.5–4 mg/ml) or PHA (5 μg/ml). HSA was used as a negative control for INA-HSA.

PBMC responses to antigens are normally expressed as a stimulation index (i.e., mean response in antigen-treated cultures/mean response in cultures containing medium alone). Thus, experiments using healthy donor cells were performed to establish 3 sigma values (i.e., mean ± 3 SD; where 99.7% of values are expected to fall within this range) for each drug treatment. This allowed us to set a reliable cut-off to discriminate between a positive and negative response in patients with liver and skin injury.

Generation of T-cell clones from patients with ATDs-induced liver injury

In initial experiments PBMC (1 × 106/well; 0.5 ml) from 2 patients [1 with ATD-related liver injury (patient TB034); 1 with ATD-related skin reaction (patient TB008)] with positive LTT or ELIspot data against at least 2 drugs were cultured with non-toxic concentrations of the ATDs (INH, 1–4 mM; INA, 1 mM, RIF, 0.1 mM; ETB, 1–2 mM; PZA, 1–4 mM; INA-NHS, 1 mM) in culture medium. In a subsequent experiment, the cloning procedure was repeated using PBMC from patient TB037 who developed liver injury following INH monotherapy. The first 2 patients were selected as they presented with skin and liver reaction and generated the strongest responses in PBMC diagnostic assays. The third patient was investigated to determine whether drug-responsive T-cells are detectable in patients exposed to INH monotherapy. Cultures were supplemented with 200 IU/ml recombinant Il-2 (PeproTech, London, UK) on days 6 and 9. On day 14, the cells were cloned by serial dilution using a previously described method (Wu et al. 2007). Epstein-Barr virus (EBV) transformed B-cell lines were created from PBMC by transformation with supernatant from the virus-producing cell line B9.58. Clones do not have to be separated from allogeneic PBMC prior to analysis as after irradiation, the cells die during the 14 day culture period. Thus, when clones are collected for testing they are 100% CD3+. Furthermore, routine screening for expression of TCR Vβeta’s reveal that clones express a single Vβeta. Thus, T-cells from the allogeneic PBMC do not contribute to the response.

Phenotype and specificity of T-cell clones

Drug specificity was assessed by culturing T-cell clones (5  ×  104 cells/well; 200 μl) with autologous antigen-resenting cells (1 × 104 cells/well) and either INH, RIF, ETB, PZA, or INA-NHS for 48 h. Previous studies have shown that 5 × 104 cells/well cultured with drug at a 5:1 ratio with antigen-presenting cells yields optimal results. Proliferation was measured by the addition of [3H]thymidine (0.5 μCi/well, 5 Ci/mmol, Morovek Biochemicals Ltd., Brea, California) for the last 16 h of culture followed by scintillation counting. Clones with a stimulation index of greater than 1.8 were expanded by repetitive stimulation with irradiated allogeneic PBMC (5  ×  104 cells/well; 200 μl) and 5 μg/ml PHA in Il-2 containing medium (250 IU/ml). Once drug-responsive clones were expanded in sufficient numbers dose-dependent reactivity against INH, RIF, ETB, PZA, INA, INA-NHS, or INA-HSA conjugates was assessed by using [3H]thymidine. T-cell phenotyping was performed by flow cytometry on a BD FACS Canto II using CD4-APC and CD8-PE antibodies (BD Biosciences). Drug-specificity of the selected T-cell clones was also assessed by ELIspot for IFN-γ, Il-5, Il-13, granzyme B, perforin, Fas Ligand, Il-10, Il-17, and Il-22 (Mabtech). Preliminary experiments reveal that the irradiated antigen-presenting cells do not produce cytokines when cultured with drug in the absence of the T-cell clones.

Mechanisms of isoniazid-specific T-cell activation

A step-wise approach was adopted to study pathways of drug-specific T-cell activation. First, T-cell clones were cultured with INH in the absence of antigen-presenting cells. Second, antigen-presenting cells were treated with MHC blocking antibodies (5 μl; BD Biosciences, Oxford, UK). Third, antigen-presenting cells pulsed with INH for 1, 4, and 24 h prior to repeat washing to remove unbound drug were used as a source of antigen. Fourth, T-cell clones, antigen-presenting cells, and INH were cultured together for 1, 2, 4, or 16 h prior to assessment of IFN-γ release by ELIspot. Cell culture supernatant was collected at each time point to measure INH albumin adducts as detailed above.

RESULTS

Drug-Specific Response of Peripheral Blood Mononuclear Cells from Patients with Liver or Skin Injury

To investigate whether drug-specific activation of T-cells is only observed in patients with liver (and skin) injury, 3 sigma values (i.e., the value where 99.7% of healthy donor responses fall within) were established for each drug. To calculate these values, PBMC from 8 drug-naïve healthy donors were cultured with optimized concentrations of INH (1 mM), ETB (1 mM), PZA (1 mM), and RIF (50 µM) and proliferative responses were measured through the addition of [3H]thymidine. Stimulation index values associated with drug treatment were as follows (mean ± SD; upper value, lower value): INH, 0.66 ± 0.35; 1.31, 0.33: ETB, 0.72 ± 0.17; 1.08, 0.54: PZA, 0.65 ± 0.38; 1.46, 0.46: RIF, 0.52 ± 0.19; 0.87, 0.45. This equates to the following upper 3 sigma values for each drug: INH, 1.72; ETB, 1.25; PZA, 1.79; RIF, 1.1. Based on this analysis, stimulation index values of 1.8 and 3 were selected to categorize weak and strong responses to the drugs in patients with ATD-related liver and skin injury.

Lymphocyte responses against ATDs were detected all 6 patients with liver or skin reactions (Table 1) (LTT or IFN-γ ELIspot Stimulation index > 1.8). PBMC from 4 patients were stimulated to proliferate with INH. Proliferative responses were also seen with INA in 1 patient. IFN-γ ELIspot responses were detected against INH, ETB and INA with PBMC from 3, 2, and 1 patient, respectively (Table 1). Patients 008, 022, 031, and 034 displayed PBMC responses (proliferation or IFN-γ secretion) against at least 2 drug antigens. Patient 043 only displayed a weak response to ETB in the ELIspot assay. It is possible that the frequency of drug-responsive T-cells in this donor was lower. Alternatively, the urticarial skin reaction might be mediated via a different mechanism. PBMC responses (proliferation and cytokine release) to INH, PZA, ETB, and RIF were not detected with patients exposed to ATDs with no signs of an adverse reaction (SI less than 1.8).

Modification of HSA by Isoniazid and Analysis of the Response of Peripheral Blood Mononuclear Cells from Patients with Liver or Skin Injury

INA-NHS was synthesized by the reaction of isonicotinoyl chloride with N-hydroxysuccinimide as previously described (Metushi et al. 2012) and characterized by NMR analysis (

A). LC-MS/MS analysis of INA-HSA revealed that the drug protein adduct was the same structure as detected in plasma of patients administered INH (B and C) (Meng et al. 2015). INA-NHS and INA-HSA did not stimulate patient PBMC to proliferate or secrete IFN-γ.

Generation of Drug-Specific T-Cell Clones

Almost 2500 T-cell clones were generated from patients 008 and 034, with ATB-related skin and liver injury, respectively, following PBMC stimulation with INH, RIF, ETB, PZA, INA, or INA-NHS. The experiment was then repeated with PBMC from liver injury patient TB037 to investigate whether similar drug-specific clones circulate in patients administered INH monotherapy. A total of 49 clones were found to display reactivity against INH (stimulation index 1.8 or more) on initial testing (Fig. 1; Table 3) and in dose-response studies following mitogen-driven expansion (Fig. 2A). Clones displaying reactivity against RIF (n = 3) and ETB (n = 1) were detected on initial testing comparing proliferative responses in medium control and 1 drug concentration (Fig. 1); however, following expansion drug-specific responses were not detected in dose–response studies and cytokine release assays. Thus, it is highly likely that the initial data are false positives. PZA, INA and INA-NHS responsive clones were not detected. Flow cytometric analysis of 39 INH clones revealed that all but one expressed the CD4+ co-receptor. The final clone expressed high levels of CD4+ and CD8+.

FIG. 1

Activation of T-cell clones with ATDs. Three thousand and twenty two clones, isolated and expanded from INH, RIF, PZA or ETB cultured PBMC of 3 patients, were cultured with antigen-presenting cells and ATDs for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Experiments were performed in duplicate. Results for each clone are expressed as a stimulation index (cpm in drug-treated wells/cpm wells containing medium alone).

FIG. 1

Activation of T-cell clones with ATDs. Three thousand and twenty two clones, isolated and expanded from INH, RIF, PZA or ETB cultured PBMC of 3 patients, were cultured with antigen-presenting cells and ATDs for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Experiments were performed in duplicate. Results for each clone are expressed as a stimulation index (cpm in drug-treated wells/cpm wells containing medium alone).

FIG. 2

Cross-reactivity of INH-responsive clones with other ATDs, INA-HSA, or INA-NHS. (A) Clones were cultured with antigen-presenting cells and titrated concentrations of ATDs for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Representative results from 4 clones are shown. (B) INA-NHS/INA-HSA conjugates did not activate INH-responsive clones. Experiments were performed in triplicate (INH, 1 mM; INA-HSA conjugate, 1 mg/ml; INA-NHS, 1 mM). (C) 1-aminobenzotriazole does not inhibit the activation of INH-specific clones. 1-Aminobenzotriazole was added 30 min prior to INH to inhibit metabolizing enzyme activity.

FIG. 2

Cross-reactivity of INH-responsive clones with other ATDs, INA-HSA, or INA-NHS. (A) Clones were cultured with antigen-presenting cells and titrated concentrations of ATDs for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Representative results from 4 clones are shown. (B) INA-NHS/INA-HSA conjugates did not activate INH-responsive clones. Experiments were performed in triplicate (INH, 1 mM; INA-HSA conjugate, 1 mg/ml; INA-NHS, 1 mM). (C) 1-aminobenzotriazole does not inhibit the activation of INH-specific clones. 1-Aminobenzotriazole was added 30 min prior to INH to inhibit metabolizing enzyme activity.

Table 3

Origin, Phenotype, and Specificity of T-Cell Clones from Patients

Patient ID Drug Clones Tested Specific Clones Phenotype (n)
 
(n(nCD4 CD8 CD4+CD8+ 
008 INH 304 
 RIF 143    
 PZA 168    
 ETB 175    
 INA-NHS 168    
034 INH 360 29 29 
 RIF 360    
 PZA 312    
 ETB 264    
 INA-NHS 208    
037 INH 576 15 
 RIF 264    
 PZA 288    
 ETB 168    
 INA-NHS 44    
Patient ID Drug Clones Tested Specific Clones Phenotype (n)
 
(n(nCD4 CD8 CD4+CD8+ 
008 INH 304 
 RIF 143    
 PZA 168    
 ETB 175    
 INA-NHS 168    
034 INH 360 29 29 
 RIF 360    
 PZA 312    
 ETB 264    
 INA-NHS 208    
037 INH 576 15 
 RIF 264    
 PZA 288    
 ETB 168    
 INA-NHS 44    

Collectively, these data confirm the presence of INH-responsive T-cells in each patient studied. Thirteen well-growing clones (3 from skin injury patient TB008 and 10 from liver injury patient TB034) were selected for subsequent experiments to characterize antigen-specificity, mechanisms of T-cell activation and the profile of cytokines secreted following drug stimulation.

Specificity of Isoniazid-Responsive CD4+ T-Cell Clones

All INH-responsive clones were stimulated to proliferate with INH in a dose-dependent manner. Reactivity against the other ATDs was not detected (Fig. 2A). Neither INA-NHS nor INA-HSA activated the clones (Fig. 2B).

1-Aminobenzotriazole Does Not Block Isoniazid-Specific T-Cell Responses

CYP-inhibitor 1-ABT, did not attenuate INH-specific T cell activation (Fig. 2C), which indicates that reactive intermediates generated in situ through INH metabolism do not activate the T-cells.

Isoniazid Activates T-Cell Clones Via 2 Pathways

Activation of T-cell clones with INH required the presence of antigen-presenting cells (Fig. 3A). Furthermore, INH-specific proliferative responses were inhibited with an anti-HLA class II, but not a class I, blocking antibody, suggesting that a drug or drug-derived antigen interacts with MHC class II molecules to activate the CD4+ clones (Fig. 3B).

FIG. 3

Antigen-presenting cell and MHC class II dependent activation of T-cell clones with INH and INH-pulsed antigen-presenting cells. (A) Activation of T cell clones required the presence of antigen-presenting cells. (B) INH-specific activation of clones was inhibited with an anti-MHC class II antibody. (C) Thirteen clones were cultured with INH-pulsed antigen-presenting cells (1–16 h; 1 mM INH) for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Data shows the mean of triplicate wells.

FIG. 3

Antigen-presenting cell and MHC class II dependent activation of T-cell clones with INH and INH-pulsed antigen-presenting cells. (A) Activation of T cell clones required the presence of antigen-presenting cells. (B) INH-specific activation of clones was inhibited with an anti-MHC class II antibody. (C) Thirteen clones were cultured with INH-pulsed antigen-presenting cells (1–16 h; 1 mM INH) for 48 h. Proliferative responses were analyzed by addition of [3H]thymidine for the final 16 h of the experiment. Data shows the mean of triplicate wells.

We have recently shown that auto-oxidation of INH leads to the formation of isonicotinic-lysine adducts on proteins (27). Such adducts will be subjected to protein processing, which may generate antigenic peptides for INH-specific T-cells. To investigate mechanisms of INH-specific T-cell activation, antigen-presenting cells were pulsed with the drug for 1, 4, and 24 h, prior to washing and exposure to clones. One out of 13 clones was activated by antigen-presenting cells pulsed with INH and the strength of the response increased with increasing length of drug exposure (Fig. 3C). The antigen-presenting cells pulsed with INH for 24 h stimulated a proliferative response similar to that seen with soluble drug (results not shown). To further investigate time-dependent T-cell activation, clones were cultured with antigen-presenting cells and soluble INH for 1–16 h, prior to washing and analysis of IFN-γ release. Activation of the antigen-presenting cell pulse positive clone was delayed, with initial and maximal IFN-γ release detected after 4 and 16 h (Fig. 4A), respectively. INH HSA adducts were measured throughout the cell culture period to characterize the level of cell exposure to protein adducts. Adducts were only detectable after 24 h incubation of 1 mM INH with EBV cells. However, when higher concentrations of INH were used, adducts were detectable after 1 h incubation (Fig. 4B). Adduct formation was time-dependent at 5 sites of modification (Lys20, 137, 195, 199, and 525) (Fig. 4B).

FIG. 4

Time-dependent IFN-γ release from INH-treated T-cell clones. (A) Clones were incubated with antigen-presenting cells and INH (1 mM) for 1, 2, 4, and 16 h. IFN-γ was measured sing ELIspot. (B) Mass spectrometry was used to qualify the time-dependent binding of INH to HSA under cell culture conditions. The epitope profile shows the lysine residues of albumin modified with INH haptens after 1–16 h.

FIG. 4

Time-dependent IFN-γ release from INH-treated T-cell clones. (A) Clones were incubated with antigen-presenting cells and INH (1 mM) for 1, 2, 4, and 16 h. IFN-γ was measured sing ELIspot. (B) Mass spectrometry was used to qualify the time-dependent binding of INH to HSA under cell culture conditions. The epitope profile shows the lysine residues of albumin modified with INH haptens after 1–16 h.

Twelve INH responsive clones were not activated when antigen-presenting cells pulsed with INH for 1–24 h were used as a source of antigen (Fig. 3C). Three of these clones were used in the INH antigen-presenting cell T-cell co-culture experiment. IFN-γ release was detected from each clone at the earliest time-point (1 h; Fig. 4A).

Secretion of Cytokines and Cytolytic Molecules from Isoniazid-Specific T-Cell Clones

ELIspot was used to screen the profile of cytokines/cytolytic molecules secreted from T-cell clones following INH treatment. All clones secreted high levels of Th1 and Th2 cytokines as well as the regulatory cytokine Il-10 and cytolytic molecules perforin, granzyme B, and FasL (Fig. 5). In contrast, Il-17 was not detected and Il-22 release was seen with only 1 clone.

FIG. 5

Secretion of cytokines and cytolytic molecules from INH-specific T-cell clones. IFN-γ, Il-5, Il-13, granzyme B, perforin, FasL, Il-10, Il-17, and Il-22 were measured by ELIspot. Clones were cultured in duplicate with antigen-presenting cells in the presence and absence of INH.

FIG. 5

Secretion of cytokines and cytolytic molecules from INH-specific T-cell clones. IFN-γ, Il-5, Il-13, granzyme B, perforin, FasL, Il-10, Il-17, and Il-22 were measured by ELIspot. Clones were cultured in duplicate with antigen-presenting cells in the presence and absence of INH.

DISCUSSION

Liver and skin injury are common and sometimes serious manifestations of ATD exposure (Galagan 2015; Kaona et al. 2004; Wares et al. 2003) that contributes to non-adherence and eventually treatment failure or relapse. Herein, we describe the characterization of INH-responsive CD4+ T-cells from patients with liver and skin injury. The cells were activated to proliferate and secrete cytotoxic mediators, and effector and regulatory cytokines, with isoniazid, but not other drugs the patients were exposed to at the time of the adverse event.

Formation of reactive intermediates by drug metabolizing enzymes results in cell stress, mitochondrial damage and/or activation of death signaling pathways that could all contribute to the tissue damage seen in patients with ATD-related adverse events. An alternative hypothesis is that the adaptive immune system is inadvertently activated following ATD exposure in patients that go on to develop tissue injury. Early studies by Warrington et al. (1978, 1982) detected drug-specific proliferation of PBMC from certain INH treated patients with liver injury. Moreover, Metushi et al. (2014a) identified anti-drug and anti-CYP P450 antibodies. Notwithstanding these investigations, the functionality of drug-specific T-cells in patients administered INH monotherapy or combination therapies has not been defined. Thus, this study focused on patients administered combinations of ATD and INH alone to explore whether drug-specific T-cell responses were detectable and if so characterize cellular phenotype/function and mechanisms of antigen presentation.

Drug-specific T-cell responses were detected in all 5 patients with ATD-related liver injury when PBMC were exposed to INH. Responses were also detected with PZA and ETB in a smaller number of patients. In most cases, the proliferation response was weak (i.e., stimulation index less than 3). Furthermore, with the exception of patients 008 and 034, drug-specific lymphocyte activation was detected with only one assay (LTT or IFN-γ ELIspot).Interestingly, the strongest responses were detected with PBMC from Caucasian patient TB034, the only patient expressing the known risk allele HLA DQB1*02:01 (Sharma et al. 2002). T-cell responses were not detected in patients exposed to ATD without adverse effects. The early study by Warrington et al. (1982) assessing lymphocyte responses in individuals given isoniazid monotherapy reported on 3 datasets categorized according to disease severity; (1) patients with severe liver failure, (2) patients with persistent liver enzyme elevation and (3) patients with transient liver enzyme elevation. INH-specific lymphocyte proliferative responses (stimulation index 2.5–8.9) were detected in 6/8 patients in group 1. In contrast, lymphocyte activation was only seen with PBMC from 4 out of 18 patients in groups 2 and 3, and with 3 of the responders the stimulation index was less than 2.9. Patients with liver injury in our cohort reside in group 2 or 3 of the Warrington classification.

These data prompted us to conduct a detailed search for INH, RIF, ETB and/or PZA-specific T-cells in patients with ATD-induced liver and skin injury by T-cell cloning. Initially, T-cell lines were generated from 2 patients (TB034—liver injury, TB008—skin injury; both exposed to 4 ATDs) by culturing PBMC individually with the 4 drugs to enrich the number of drug-specific T-cells. Almost 2500 clones were generated by serial dilution and repetitive mitogen-driven activation. The protocol was subsequently repeated with blood mononuclear cells from patient TB037 who developed liver injury following INH monotherapy. Initial screening (drug at a single concentration versus control in duplicate) identified over 50 INH-responsive clones (from all patients), 3 RIF-responsive clones (from patient TB008) and 1 ETB-responsive clone (from patient TB034). However, following expansion and repeated testing with titrated drug concentrations and using cytokine release assays, T-cell activation was only detected with INH. Forty-nine CD4+ clones originating from all 3 patients displayed robust and reproducible INH-specific proliferative responses in a dose-dependent manner. The clones were not activated with other ATDs, which is what one would expect to find with such structurally different antigenic determinants. The inability to obtain clones responsive for RIF, ETB and PZA from patients with liver and skin reactions indicate that INH is the only drug that activates a T-cell response in this cohort of patients. If future studies reveal that T-cells are responsible for the tissue injury, INH seems to be the only implicated drug. However, at this point we cannot exclude the possibility that RIF, ETB, and PZA exert other direct effects on cell function and/or activate the innate immune system.

Activation of the clones with INH was dependent on antigen-presenting cells and proliferative responses were inhibited with an MHC class II blocking antibody. CD4+ T cells are known to mediate immunological diseases through the release of cytokines that promote antibody class switching in B-cells, enhance the cytolytic activity of CD8+ T cells and regulate the strength and duration of the immune response. Furthermore, CD4+ T-cells can cause tissue injury directly through the release of cytolytic mediators (Chen et al 1990; Jacobson et al. 1984; Tite and Janeway 1984). To execute these important functions, they differentiate into unique subsets characterized by the profile of specific cytokines released following T-cell receptor triggering (Marshall and Swain 2011). In recent years, new T-cell populations have been identified that secrete cytokines such as Il-17 and Il-22, which mediate inflammatory conditions through an interaction with receptors that reside in specific tissues. INH-specific clones secreted IFN-γ, Il-5, and Il-13 alongside cytotoxic mediators perforin, granzyme B, and FasL, indicating that they might damage target cells directly. Il-17 was not detected, while Il-22 secretion was seen with only one clone. In contrast to antigen-specific T-cell clones isolated from patients with other forms of immunological drug reaction (Bell et al. 2013; El-Ghaiesh et al. 2012; Wu et al. 2007), high levels of Il-10 was secreted from all clones following activation with INH. In a prospective analysis of 35 patients undergoing INH therapy, Metushi et al. (2014c) found an increased frequency of Il-17 and Il-10 secreting cells in patients with mild elevations in liver enzymes through the surveillance of PBMC by flow cytometry. Our data indicates that the increase in Il-10 secreting cells likely derives from the expansion of drug-specific T-cells and that these cells help to regulate the immune response and subsequently the extent of tissue damage, effectively preventing acute liver failure (and possibly severe skin reactions). In contrast, if Il-17 secreting cells participate in ATD-induced reactions, their expansion most likely results from non-antigen specific bystander activation as has been reported previously in patients with allergic contact dermatitis (Pennino et al. 2010).

Drugs activate T-cells through (1) direct reversible modification of MHC molecules and/or specific T-cell receptors (PI concept [e.g., sulfamethoxazole, carbamazepine, lamotrigine]; altered peptide repertoire concept [e.g., abacavir]); (2) direct irreversible modification of MHC or MHC binding peptides (e.g., nitroso sulfamethoxazole); and (3) irreversible modification of non-MHC proteins liberating neo-antigens through protein processing (e.g., β-lactam antibiotics). Interestingly, multiple pathways of drug-specific T-cell activation occur in the same patient. This phenomenon has clearly been demonstrated in patients with reactions to β-lactam antibiotics, sulfonamides, and abacavir (Uetrecht and Naisbitt, 2013). INH is metabolized to structurally diverse intermediates (acetylhydrazine, INA, isonicotinic amide, acyl diazene, diazohydroxide (Li et al. 2011; Mahapatra et al. 2012; Metushi et al. 2012), which have the potential to bind irreversibly to protein. Moreover, the detection of anti-INH antibodies in patients with liver injury by Metushi et al. (2014a) indicates that the adaptive immune system is activated with INH protein adducts. However, 1-aminobenzotriazole, which inhibits metabolic activity in immune cells (Sanderson et al. 2007), did not block the activation of clones with INH. Moreover, T-cell proliferative responses were not detected with INA. Collectively, these data suggest that the clones are not activated with an INH intermediate generated in situ through metabolism in the T-cell assay.

INH has also been shown to bind irreversibly to protein via an auto-oxidation pathway (Meng et al. 2015). Thus, to explore mechanisms of INH-specific T-cell activation, kinetic assays were conducted where (1) antigen-presenting cells pulsed with INH for 1–16 h were used as a source of antigen and (2) antigen-presenting cells, T-cells and INH were cultured together for 1–16 h prior to analysis of IFN-γ release. At each stage of the analysis INH albumin adducts were measured by mass spectrometry to confirm that adducts are formed in cell culture medium via auto-oxidation. Most of the INH-responsive clones were not stimulated to proliferate with INH pulsed antigen-presenting cells, which suggests T-cell activation occurs through a direct reversible interaction between drug and the immunological receptors. Moreover, with these clones, T-cell activation was rapid. IFN-γ release was detected after 1 h, which is unlikely to occur when the formation of protein adducts and protein processing is required to generate antigenic determinants. In contrast, one clone was activated by antigen-presenting cells pulsed with INH for 4–16 h, the time required to generate high quantities of protein adducts. Furthermore, when antigen-presenting cells, T-cells and INH were cultured together, IFN-γ release was delayed for a similar duration. This clone is probably activated via a hapten mechanism, the antigen being a peptide derived from a protein adduct generated through INH auto-oxidation. One should note that the clones described herein were isolated from PBMC cultured in the presence of the parent compound. Hence, it is not surprising that the majority were responsive to INH itself. Further studies are required to characterize the INH-derived adducts formed in antigen-presenting cells (or liver cells) that act as T-cell antigens before it is possible to comment on whether the parent drug, protein adducts or both drive T-cell responses in patients with liver and skin injury.

In conclusion, our data shows that INH-specific T-cells are clearly detectable in patients with ATD-mediated liver and skin injury. In contrast, responses were not detected in drug tolerant controls. T-cells were activated through a direct reversible interaction of INH with immune cells and via a hapten mechanism. In ongoing experiments, we are studying the nature of the T-cell response prospectively when patients develop acute reactions and in patients with severe liver failure.

SUPPLEMENTARY DATA

Supplementary data are available at Toxicological Sciences online.

ACKNOWLEDGMENTS

The authors thank the patients for their generous blood donation.

FUNDING

This work received core-funding from the Medical Research Council Centre for Drug Safety Science (Grant Number MR/L006758/1). Part-funded was obtained from the MIP-DILI project which is supported by the European Community under the Innovative Medicines Initiative (IMI) Programme through Grant Agreement number 115336. TU is a visiting scientist at the MRC Centre for Drug Safety Science from Sumitomo Dainippon Pharma Co.,LTD.

REFERENCES

Agal
S.
Baijal
R.
Pramanik
S.
Patel
N.
Gupte
P.
Kamani
P.
, et al.  . (
2005
).
Monitoring and management of antituberculosis drug induced hepatotoxicity
.
J. Gastroenterol. Hepatol.
 
20
,
1745
1752
.
Bell
C. C.
Faulkner
L.
Martinsson
K.
Farrell
J.
Alfirevic
A.
Tugwood
J.
, et al.  . (
2013
).
T-cells from HLA-B*57:01+ human subjects are activated with abacavir through two independent pathways and induce cell death by multiple mechanisms
.
Chem. Res. Toxicol.
 
26
,
759
766
.
Chen
B. P.
Madrigal
A.
Parham
P.
(
1990
).
Cytotoxic T cell recognition of an endogenous class I HLA peptide presented by a class II HLA molecule
.
J. Exp. Med.
 
172
,
779
788
.
Chowdhury
A.
Santra
A.
Bhattacharjee
K.
Ghatak
S.
Saha
D. R.
Dhali
G. K.
(
2006
).
Mitochondrial oxidative stress and permeability transition in isoniazid and rifampicin induced liver injury in mice
.
J. Hepatol.
 
45
,
117
126
.
El-Ghaiesh
S.
Monshi
M.
Whitaker
P.
Jenkins
R.
Meng
X.
Farrell
J.
, et al.  . (
2012
).
Characterization of the antigen specificity of T-cell clones from piperacillin hypersensitive patients with cystic fibrosis
.
J. Pharmacol. Exp. Ther.
 
341
,
597
610
.
Galagan
J. E.
(
2014
).
Genomic insights into tuberculosis
.
Nat. Rev. Genet.
 
15
,
307
320
.
Jacobson
S.
Richert
J. R.
Biddison
W. E.
Satinsky
A.
Hartzman
R. J.
McFarland
H. F.
(
1984
).
Measles virus-specific T4+ human cytotoxic T cell clones are restricted by class II HLA antigens
.
J. Immunol.
 
133
,
754
757
.
Kaona
F. A.
Tuba
M.
Siziya
S.
Sikaona
L.
(
2004
).
An assessment of factors contributing to treatment adherence and knowledge of TB transmission among patients on TB treatment
.
BMC Public Health
 
4
,
68.
Lehloenya
R. J.
Muloiwa
R.
Dlamini
S.
Gantsho
N.
Todd
G.
Dheda
K.
(
2015
).
J Lack of cross-toxicity between isoniazid and ethionamide in severe cutaneous adverse drug reactions: a series of 25 consecutive confirmed cases
.
Antimicrob Chemother
 .
70
,
2648
2651
.
Li
F.
Miao
Y.
Zhang
L.
Neuenswander
S. A.
Douglas
J. T.
Ma
X.
(
2011
).
Metabolomic analysis reveals novel isoniazid metabolites and hydrazones in human urine
.
Drug Metab. Pharmacokinet.
 
26
,
569
576
.
Marshall
N. B.
Swain
S. L.
(
2011
).
Cytotoxic CD4 T cells in antiviral immunity
.
J. Biomed. Biotechnol.
 
2011
,
954602.
Mahapatra
S.
Woolhiser
L. K.
Lenaerts
A. J.
Johnson
J. L.
Eisenach
K. D.
Joloba
M. L.
, et al.  . (
2012
).
A novel metabolite of antituberculosis therapy demonstrates host activation of isoniazid and formation of the isoniazid-NAD+ adduct
.
Antimicrob. Agents Chemother.
 
56
,
28
35
.
Meng
X.
Maggs
J. L.
Usui
T.
Whitaker
P.
French
N. S.
Naisbitt
D. J.
, et al.  . (
2015
).
Auto-oxidation of Isoniazid Leads to Isonicotinic-Lysine Adducts on Human Serum Albumin
.
Chem. Res. Toxicol.
 
28
,
51
58
.
Metushi
I. G.
Nakagawa
T.
Uetrecht
J.
(
2012
).
Direct oxidation and covalent binding of isoniazid to rodent liver and human hepatic microsomes: humans are more like mice than rats
.
Chem. Res. Toxicol.
 
25
,
2567
2576
.
Metushi
I. G.
Sanders
C.
Lee
W. M.
Uetrecht
J.
(
2014a
).
Detection of anti-isoniazid and anti-cytochrome P450 antibodies in patients with isoniazid-induced liver failure
.
Hepatology
 
59
,
1084
1093
.
Metushi
I. G.
Lee
W. M.
Uetrecht
J.
(
2014b
).
IgG3 is the dominant subtype of anti-isoniazid antibodies in patients with isoniazid-induced liver failure
.
Chem. Res. Toxicol.
 
27
,
738
740
.
Metushi
I. G.
Zhu
X.
Chen
X.
Gardam
M. A.
Uetrecht
J.
(
2014c
).
Mild isoniazid-induced liver injury in humans is associated with an increase in Th17 cells and T cells producing IL-10
.
Chem. Res. Toxicol.
 
27
,
683
689
.
Pennino
D.
Eyerich
K.
Scarponi
C.
Carbone
T.
Eyerich
S.
Nasorri
F.
, et al.  . (
2010
).
IL-17 amplifies human contact hypersensitivity by licensing hapten nonspecific Th1 cells to kill autologous keratinocytes
.
J. Immunol.
 
184
,
4880
4888
.
Ramappa
V.
Aithal
G. P.
(
2013
).
Hepatotoxicity related to anti-tuberculosis drugs: mechansisms and management
.
J. Clin. Exp. Hepatol
 .
3
,
37
49
.
Sanderson
J. P.
Naisbitt
D. J.
Farrell
J.
Ashby
C. A.
Tucker
M. J.
Rieder
M. J.
, et al.  . (
2007
).
Sulfamethoxazole and its metabolite nitroso sulfamethoxazole stimulate dendritic cell costimulatory signaling
.
J. Immunol.
 
178
,
5533
5542
.
Sharma
S. K.
Balamurugan
A.
Saha
P. K.
Pandey
R. M.
Mehra
N. K.
(
2002
).
Evaluation of clinical and immunogenetic risk factors for the development of hepatotoxicity during antituberculosis treatment
.
Am. J. Respir. Crit. Care Med.
 
166
,
916
919
.
Tite
J. P.
Janeway
C. A.
Jr.
(
1984
).
Cloned helper T cells can kill B lymphoma cells in the presence of specific antigen: Ia restriction and cognate vs. noncognate interactions in cytolysis
.
Eur. J. Immunol.
 
14
,
878
886
.
Uetrecht
J.
Naisbitt
D. J.
(
2013
).
Idiosyncratic adverse drug reactions: current concepts
.
Pharmacol Rev
 .
65
,
779
808
.
Wares
D. F.
Singh
S.
Acharya
A. K.
Dangi
R.
(
2003
).
Non-adherence to tuberculosis treatment in the eastern Tarai of Nepal
.
Int. J. Tuberc. Lung Dis.
 
7
,
327
335
.
Warrington
R. J.
Tse
K. S.
Gorski
B. A.
Schwenk
R.
Sehon
A. H.
(
1978
).
Evaluation of isoniazid-associated hepatitis by immunological tests
.
Clin. Exp. Immunol.
 
32
,
97
104
.
Warrington
R. J.
McPhilips-Feener
S.
Rutherford
W. J.
(
1982
).
The predictive value of the lymphocyte transformation test in isoniazid-associated hepatitis
.
Clin. Allergy
 
12
,
217
222
.
Whitaker
P.
Meng
X.
Lavergne
S. N.
El-Ghaiesh
S.
Monshi
M.
Earnshaw
C.
, et al.  . (
2011
).
Mass spectrometric characterization of circulating and functional antigens derived from piperacillin in patients with cystic fibrosis
.
J. Immunol.
 
187
,
200
211
.
Wu
Y.
Farrell
J.
Pirmohamed
M.
Park
B. K.
Naisbitt
D. J.
(
2007
).
Generation and characterization of antigen-specific CD4+, CD8+, and CD4+CD8+ T-cell clones from patients with carbamazepine hypersensitivity
.
J. Allergy Clin. Immunol.
 
119
,
973
981
.

Supplementary data