CLONAL REPERTOIRE ANALYSIS IN DRUG HYPERSENSITIVITY

CLAIM OF PRIORITY

This application claims the benefit of U.S. Provisional Application Serial No. 63/381,946, filed on November 2, 2022. The entire contents of the foregoing are incorporated herein by reference.

FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

This invention was made with Government support under Grant No. OD023091 awarded by the National Institutes of Health. The Government has certain rights in the invention.

TECHNICAL FIELD

Provided herein are assays and methods for identifying which of a plurality of drugs is responsible for causing a delayed-type drug hypersensitivity reaction (dtDHR) in a patient.

BACKGROUND

Delayed-type drug hypersensitivity reactions (dtDHR) are a significant cause of morbidity and mortality with considerable cost to healthcare systems (1-4). Skin is the most commonly affected organ and severity ranges from a mild skin-limited reaction (morbilliform drug eruption, MDE, also known as maculopapular exanthem or MPE) to life-threatening severe cutaneous adverse reactions (SCAR) which have skin and systemic inflammation. The most severe forms of SCAR are Stevens-Johnson syndrome/toxic epidermal necrolysis (SJS/TEN), notable for blistering and sloughing of skin and mucosal tissues requiring bum unit level care, and drug reaction with eosinophilia and systemic symptoms (DRESS), notable for potentially severe visceral involvement. There is currently no reliable test to identify culprit drug or to screen patients to prevent a reaction to drug. This limitation is a major impediment to improving clinical care of dtDHR patients. SUMMARY

Provided herein is a method of determining which of a plurality of drugs to which a patient has had or is having a Type 4 allergic reaction, comprising: (a) obtaining a sample from the patient and culturing a portion of the sample with each of the plurality of drugs suspected of causing or having caused the allergic reaction as a test sample, and culturing another portion of the sample without any of the plurality of drugs as a control sample; (b) subjecting the test sample and the control sample to T cell receptor (TCR) sequencing; (c) detecting clonal TCR expansion to detect the presence of expanded clones in the test sample exposed to the drug as compared to non-expanded clones; and (d) identifying the drug in the test sample having clonal TCR expansion as being the drug to which the patient is having or has had an allergic reaction.

Also provided herein are methods comprising: (a) obtaining a sample from the patient who has had or is having a Type 4 allergic reaction, and who has been administered a plurality of drugs, one of which is suspected of causing or having caused the allergic reaction, and culturing a portion of the sample with each of the plurality of drugs suspected of causing or having caused the allergic reaction as a test sample, and culturing another portion of the sample without any of the plurality of drugs as a control sample; (b) subjecting the test sample and the control sample to T cell receptor (TCR) sequencing; (c) detecting clonal TCR expansion to detect the presence of expanded clones in the test sample exposed to the drug as compared to non-expanded clones; and (d) identifying the drug in the test sample having clonal TCR expansion as being the drug to which the patient is having or has had an allergic reaction.

Additionally provided herein are methods of determining whether a patient is likely to have a Type 4 drug reaction to a drug the patient has not yet taken, the method comprising the steps of: (a) obtaining a sample from the patient and culturing a portion of the sample with the drug as a test sample, and culturing another portion of the sample without the drug as a control sample; (b) subjecting the test sample and the control sample to T cell receptor (TCR) sequencing; (c) detecting clonal TCR expansion to identify expanded clones in the test sample exposed to the drug as compared to nonexpanded clones; and (d) if the test sample has clonal TCR expansion above a first threshold, identifying the drug as being a drug that is likely to cause a Type 4 allergic reaction and should not be administered; or if the test sample does not have clonal TCR expansion, identifying the drug as being a drug that is less likely to cause a Type 4 allergic reaction and can be administered. In some embodiments, if the test sample has clonal TCR expansion below the first threshold but above a second, threshold, identifying the drug as being a drug that may cause a Type 4 allergic reaction and should only be administered with caution.

In some embodiments, the sample from the patient is a skin sample, optionally from a reaction site.

In some embodiments, the sample from the patient is a blood sample.

In some embodiments, the patient has had an allergic reaction to a drug in the past.

In some embodiments, the patient is having an allergic reaction when the sample is obtained, or wherein the patient has an allergic reaction that has resolved before the sample is obtained.

In some embodiments, one or more concentrations of the drug to be tested are used.

In some embodiments, the test and control samples are incubated, optionally for 2-7 days, and thereafter restimulated with either the drug or vehicle control, and cultured, optionally for an additional 1-4 days preferably 1-2 days, before step (b).

In some embodiments, the methods described herein comprise isolating migrating immune cells before step (b).

In some embodiments, the methods further comprise analysing residual skin tissue for cell death and/or gross loss of integrity, and/or analysing markers in supernatant from skin culture.

In some embodiments, the supernatant is analyzed for markers of T cell effector function.

In some embodiments, the markers are selected from granulysin, interferon gamma (IFNg), interleukin 5 (IL5), IL 10, and IL13.

In some embodiments, the residual skin tissue is analyzed for markers of keratinocyte death.

In some embodiments, the marker of keratinocyte death is lactate dehydrogenase (LDH).

In some embodiments, the methods further comprise enriching CD8+ T cells and/or CD14+ monocytes from PBMCs from the sample, and/or depleting CD25+ (Treg) cells from the sample, before culturing.

In some embodiments, the methods further comprise identifying the drug in the test sample as being the drug to which the patient is having or has had an allergic reaction when the test sample has two or more of TCR clonal expansion, cell death and/or loss of integrity in the skin sample, and markers in supernatant from skin culture. In some embodiments, the markers are selected from granulysin, interferon gamma (IFNg), interleukin 5 (IL5), IL 10, and IL13.

Provided herein are methods of testing a patient to determine drugs to which the patient has had or is having a Type 4 allergic reaction comprising the steps of: (a) obtaining a test sample from the patient and culturing a portion of the sample with the drugs suspected of causing or having caused the patient’s allergic reaction along with culturing another portion of the sample as a vehicle control, (b) subjecting the test sample and the control sample to TCR sequencing, and (c) analyzing the results for clonal expansion to identify expanded clones in the test sample exposed to the drug as compared to vehicle control; if clonal expansion is identified, the tested drug is the drug to which the patient is having or has had an allergic reaction.

In some embodiments, the test sample from the patient is a skin sample. In some embodiments, the test sample from the patient is a blood sample.

In some embodiments, the patient has had an allergic reaction to a drug in the past.

In some embodiments, the patient is having an allergic reaction when the test is being conducted.

Also provided herein are methods of determining whether a patient is likely to have a Type 4 drug reaction to a drug the patient has not yet taken comprising the steps of: (a) obtaining from the patient a test sample and culturing a portion of the sample with the drug intended for administration along with culturing another portion of the sample as a vehicle control, (b) subjecting the test sample and the control sample to TCR sequencing, and (c) analyzing the results for clonal expansion to identify expanded clones in the test sample exposed to the drug as compared to vehicle control; if clonal expansion is identified, the tested drug that is intended for administration is likely to cause a Type 4 allergic reaction and should not be administered.

In some embodiments, the test sample from the patient is a skin sample. In some embodiments, the test sample from the patient is a blood sample.

In some embodiments, the method is run iteratively on a series of drugs until a drug is identified to which the patient will not react.

In some embodiments, one or more concentrations of the drug to be tested are used. In some embodiments, the test and control samples are incubated for 2-4 days and thereafter restimulated with either the drug or vehicle control, and cultured for an additional 1-2 days.

In some embodiments, migrating cells, residual skin tissue and supernatant from skin culture are employed.

In some embodiments, additionally PBMCs from the patient’s blood are collected and either CD8+ T cells and CD14+ monocytes are isolated or CD25+ (Treg) cells are depleted and then the cells cultured.

In some embodiments, the sample is analyzed for markers of T cell effector function. In some embodiments, the markers are selected from granulysin, IFNg, IL5 and IL13.

In some embodiments, the sample is analyzed for markers of keratinocyte death. In some embodiments, the marker of keratinocyte death is LDH.

In some embodiments, HLA sequencing is additionally performed on the residual DNA and the TCR sequences of the clones and the HLA sequences are compared to known TCR and HLA sequences to identify potential public clones to determine if the same T cell clones were expanded in other patients with the same HLA allele and with the same drug reaction from the drug in question.

In some embodiments, the methods further comprise treating the patient or recommending a treatment decision for the patient using a method shown in Table A.

Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. Methods and materials are described herein for use in the present invention; other, suitable methods and materials known in the art can also be used. The materials, methods, and examples are illustrative only and not intended to be limiting. All publications, patent applications, patents, sequences, database entries, and other references mentioned herein are incorporated by reference in their entirety. In case of conflict, the present specification, including definitions, will control.

Other features and advantages of the invention will be apparent from the following detailed description and figures, and from the claims.

DESCRIPTION OF DRAWINGS

FIGs. 1A-E. Retrospective analysis of skin samples demonstrates variable T cell phenotypes and function across dtDHR severity. (A) Representative H&E images of each dtDHR. Black bar = 100 m. (B) Immunofluorescent staining for CD3, CD8, and CLA, with DAPI nuclear stain. White bar = 100 pm. (C) Log2counts of T cell phenotypic genes in dtDHR and healthy skin. (D) Volcano plots highlighting significantly differentially expressed functional markers in diseased versus healthy skin. (E) Log2counts of functional markers in dtDHR and healthy skin. (A,B) N = 3-6 per dtDHR. (C-E) N = 13 SJS/TEN, 6 DRESS, 6 MDE, and 11 healthy. * Significance defined as Log2FC > ±1 and P _a dj < 0.05.

FIGs. 2A-C. Prospective analysis by scRNAseq + CITEseq of dtDHR reveals complexity of T cell phenotypes mediating disease. (A) 22 T cell clusters were identified from integrating scRNAseq + CITEseq across 3 SJS/TEN, 3 MDE and 3 healthy control skin and 3 healthy control blood samples. One healthy control skin sample was excluded due to low number of cells. Healthy control blood and skin were not paired. Heatmap shows clusters by phenotypic and functional markers using both genes (italicized) and protein (bolded) markers. (B) Median percentage + range cytotoxic CD8 ⁺ T cells and Treg in skin and blood of SJS/TEN, MDE, and H.C. patients. (C) Percentage cytotoxic T cells of total T cells in individual patient skin and blood across cytotoxic clusters identified in the heatmap (A).

FIG. 3. Volcano plots highlighting selected significantly differentially expressed genes between pooled SJS/TEN, MDE, and healthy controls (H.C.), in skin and in blood. Significance defined as Log2FC > ±1 and P _a dj < 0.05.

FIGs. 4A-C. TCR sequencing suggests clonal expansion and recruitment of cytotoxic CD8 ⁺ T cells in SJS/TEN but not in MDE. (A) Clonal frequency (percentage) of the top 25 clones in skin and in blood of each dtDHR patient. (B) Bar graph showing percent distribution across T cell phenotypic clusters of the top clone in skin (grey). If that same clone was also found in blood of that patient, it is additionally shown in open sections. (C) Table showing fold change of clones in ex vivo culture of PBMCs from SJS/TEN patient 1 with suspected culprit drug, bupropion, at two concentrations compared to culture with vehicle alone. The top five clones (SEQ ID NOs: l-5) deemed expanded in blood (from A) are individually shown and color coded to match (A). Additionally the number of clones deemed not to have proliferated in ex vivo culture with drug (fold change < 1) is shown. FIGs. 5A-B. TCRseq analysis of healthy control skin samples. (A) Clonal frequency (percentage) of the top 25 clones in skin each healthy control. (B) Bar graph showing percent distribution across T cell phenotypic clusters of the top clone in skin.

FIGs. 6A-D. Human skin TRM can mediate MDE despite few circulating T cells. (A) Representative H&E images from a lymphopenic patients with MDE. Black bar = 100 pm. (B) Immunofluorescence staining for CD3, CLA, and CD8 (left image) and CD3, CD45RO, and CD45RA (right image), with DAPI nuclear stain. White bar = 100 pm. (C) Total lymphocyte count in peripheral blood in lymphopenic vs non- lymphopenic patients with MDE. Lines show median. Significance defined as p < 0.05, two-tailed Mann-Whitney test. (D) Number of CD3 ⁺ T cells quantified per high powered field (hpf) in immunofluorescently stained tissue sections in lymphopenic and non-lymphopenic patients with MDE and healthy skin. Lines show median. Not significant, p > 0.05, Kruskal -Wallis test.

FIGs. 7A-F. TRM are present in skin after disease resolution. Mice treated with systemic drug alone did not develop skin inflammation by (A) ear thickness (mean with standard deviation of the mean, (SEM)) shown or (B) total number of CD3 ⁺ T cells and CD8 ⁺ T cells in ear skin by flow cytometry. Comparatively, mice treated with systemic and topical drug develop skin inflammation that slowly resolves by 90 days post-treatment as measured by (C) ear thickness and (D) H&E analysis. (E) Total number of CD8 ⁺ TEM (CD44 ^lll8ll CD62L ^l0 " ) in blood, TCM (CD44 ^high CD62L ^high ) in lymph node and total CD8 ⁺ T cells in ear skin quantified by flow cytometry. Lines show median. Significance defined as p < 0.05, Kruskal -Wallis test followed by Dunn’s multiple comparisons test between experimental group and each control group. (F) CD8 ⁺ T cells in ear skin show a TRM (CD62L ^low CD69 ⁺ CLA ⁺ ) phenotype by flow cytometry. Plots gated on CD3 ⁺ CD8 ⁺ T cells.

FIGs. 8A-F. Skin TRM mediate MDE-like reaction in mice in the absence of circulating T cells. (A) Schematic of drug challenge experiment. (B) Ear thickness through 107 days (peak of challenge). Mean with SEM shown. (C) Representative H&E images at day 107. (D) Total number of CD8 ⁺ T cells in ear skin at day 107. Lines show median. Significance defined as p < 0.05, Kruskal-Wallis test followed by Dunn’s multiple comparisons test. Only challenge without FTY720 versus primary response was significant. (E) Percentage CD3 ⁺ T cells and CD3 ⁺ CD8 ⁺ T cell in blood, and percentage of effector CD8 ⁺ T cells (CD44 ^high CD62L ^low ) in blood at 107 days, demonstrating systemic drug challenge is associated with a systemic T cell reaction. Lines show median. Significance defined as p < 0.05, Kruskal -Wallis test followed by Dunn’s multiple comparisons test. (F) Total number and percentage of functional CD8 ⁺ T cells in ear skin of mice treated or not with FTY720. All comparisons not significant with p > 0.05 two-tailed Mann-Whitney test. (A-F) N > 3 mice per group. Each experiment was repeated at least twice.

FIGs. 9A-C. T cell receptor sequencing identifies drug-reactive clones in skin affected by fixed drug eruption (FDE). (A) Clinical pictures and associated clonal frequency (percentage) from two biopsy sites taken simultaneously. Two patients shown. Paired clones between each biopsy site are color coded. (B) Clonal frequency (percentage) in an active FDE site and a non-FDE site taken simultaneously from a third patient. (C) Clonal frequency (percentage) in an active FDE site and a non-FDE site taken simultaneously from a fourth patient. The active FDE site was re-biopsied 5 years and another 1 year after resolution.

FIGs. 10A-D. T cell receptor sequencing identifies an expanded drug-reactive clone in skin and blood. (A) Productive frequency (percentage) of T cell clones in skin across 4 samples: FFPE skin biopsy collected at the start of SJS/TEN, a research skin biopsy - half cultured with presumed culprit drug, palbociclib O.lng/ml, and half cultured with vehicle, and then a FFPE skin biopsy collected from an unrelated rash presenting weeks later. Each clone is color coded across samples. Biopsy - bx. (B) Productive frequency (percentage) of T cell clones in a research sample of PBMCs cultured with vehicle, palbociclib at 3 different concentrations or binimetinib at 2 concentrations. Each clone is color coded across samples and matches color coding in skin (A). (C) Granulysin in supernatant of skin cultures with vehicle or palbociclib. (D) Granulysin in supernatant of PBMC cultures with vehicle, palbociclib or binimetinib.

DETAILED DESCRIPTION

Immune-mediated drug reactions (drug allergies) fall under one of six types of drug reactions per the World Health Organization classification system (5). Immune- mediated drug reactions are further classified as types 1 - 4 according to the Gell and Coombs model of hypersensitivity (6). Type 4 reactions are termed delayed-type reactions, given that they typically begin days to weeks after exposure to inciting antigen. Reactions can occur more quickly, on the order of one or more days, if history of prior antigen exposure. This timing is consistent with a T cell mediated response. Indeed, T cells and molecules typically attributed to T effector cells are consistently detected in skin biopsies or blister fluid from dtDHR (7, S). Moreover, since 2002 there has been increasing recognition of HLA allele associations with specific drug:dtDHR (9-73) further supporting T cells as key protagonists in this disease. SJS/TEN research repeatedly has shown a cytotoxic CD8 ⁺ T cell predominance in skin and blister fluid (7, 14-19). However, the identity of CD8 ⁺ T cell subset(s) mediating disease, their origin, and mechanism and location of activation remain unknown. Even the putative effector function by which the CD8 ⁺ T cells mediate keratinocyte death is debated. While some studies support cytotoxic granule components or Fas:FasL (20), significant data suggest granulysin may be the primary effector molecule (7), and most recently a completely novel mechanism has been reported by which T cell activated monocytes mediate keratinocyte death (27). Data are even more limited in DRESS and MDE, which is particularly noteworthy for MDE given its prevalence. It is currently unclear whether DRESS and MDE are CD4 ⁺ or CD8 ⁺ T cell mediated and whether they are Thl, Th2 or Th 17 polarized. They are presumed to be non-cytotoxic, though data to that effect are limited.

Skin resident memory T cells (TRM) are a unique population of memory T cells that reside long-term in skin without recirculating even during periods of immunologic quiescence (22-24). Skin TRM are increasingly implicated in the pathogenesis of several inflammatory skin diseases, most notably allergic contact dermatitis (25), which is another form of delayed-type hypersensitivity reaction, and acute graft-versus-host- disease (2d), which clinically and histologically can present identically to MDE, DRESS and SJS/TEN. Subsequently, a role for skin TRM in dtDHR has been surmised. Recent research supports that skin TRM are generated by dtDHR (27), but there are no studies to date investigating whether skin TRM mediate disease. Importantly, knowing whether TRM or other T cell subsets mediate disease is a critical step in illuminating disease pathogenesis, but moreover has significant clinical implications such as development of a reliable test to identify culprit drug, which is currently lacking in the clinic.

Notably, dtDHR pathobiology is historically under-researched (4) due to three main barriers: (i) the rarity and acuity of severe disease impedes prospective sample collection; (ii) skin samples obtained for clinical purposes are typically formalin fixed paraffin embedded (FFPE) which previously precluded extensive laboratory analysis; and (iii) lack of adequate mouse models. Herein, we leveraged recent technical advancements to overcome these prior limitations to interrogate the origin, phenotype and function of pathogenic T cells, in particular, skin TRM, in dtDHR.

Results incriminated a number of T cell subsets, resident and recruited, clonally expanded and not, as potentially pathogenic in SJS/TEN. Most strikingly, SCAR were marked by systemic drug-reactive CD8 ⁺ T cell activation that was lacking in MDE. These data are supported by reports by Villani et al who observed clonal expansion in blister fluid/skin and blood in the majority of TEN patients but not MDE patients assayed by high-throughput TCR sequencing (65), and Pan et al who detected clonal expansion of drug-reactive cytotoxic CD8 ⁺ T cells in blister fluid and blood in carbamazepine-induced SJS/TEN but not carbamazepine-tolerant controls by next generation sequencing (56). Microscopy and bulk transcriptional profiling paralleled scRNAseq + CITEseq + TCRseq data in implicating skin TRM as potential protagonists in MDE, and our observations in humans and mice unable to effectively recruit T cells into skin provided functional evidence that skin-resident T cell populations are sufficient to mediate MDE. These findings make teleological sense since SCAR patients suffer from systemic involvement while MDE is skin-limited.

The identification of clonal expansion across multiple T cell subsets implies that either a drug-reactive precursor proliferated and differentiated into multiple phenotypes during this dtDHR episode, or a prior exposure generated clones of multiple phenotypes that were reactivated upon antigen re-exposure. The latter has been demonstrated in allergic contact dermatitis, another form of delayed-type hypersensitivity (25). If this is the case in dtDHR, at-risk patients could potentially be identified by testing for preexisting drug-reactive skin TRM before drug exposure. Moreover, pre-existing drug- reactive TRM in skin invokes a potential role for heterologous immunity in dtDHR pathogenesis whereby patients have previously generated virus-specific skin TRM that cross-react to drug (70-72). This could explain why in some cases patients react upon first exposure to drug. Further work is necessary to explore these possibilities.

Based on our results, we surmised that clonal repertoire analysis (T cell receptor sequencing) could be used to identify drug-reactive clones in patient skin and blood and thus could serve as a test to identify culprit drug. Indeed, clonal expansion was observed in PBMCs cultured with culprit drug but not when cultured with vehicle control. Expanded clones in culture matched clones that were expanded during active disease supporting specificity of testing. We also observed clonal expansion of drug- reactive clones in skin of active fixed drug eruption (FDE) lesions that was not observed in non-FDE lesions/control skin. Drug-reactive clones remained in skin up to 6 years after disease resolution. FDE is another CD8 mediated form of dtDHR, supporting that clonal expansion occurs in various forms of dtDHR, and again supporting specificity. Finally, we observed drug-specific clonal expansion in test skin and blood from a patient with SJS/TEN. The expanded clone matched the diseasemediating clone identified in a skin biopsy taken during active disease. Further, skin biopsy taken from an unrelated rash showed an entirely different expanded clone, again demonstrating specificity of clonal sequencing as a read-out for drug reactivity in dtDHR.

Provided herein are methods that can be used to test to what drug(s) a patient is/has reacted, i.e., to determine to what drug they are allergic, for type 4 drug reactions, not type 1 drug reactions.

The methods can include an initial expert review/interpretation of history of prior disease at a clinical and/or immunologic level based on clinical history and skin biopsy, if available.

The methods can be practiced using samples comprising skin samples, e.g., 5- 10mm, e.g., 6mm, skin biopsies (primary test target) and/or blood samples (secondary test target) from patients. The samples can be obtained after disease has resolved or during active disease. In preferred embodiments, the skin biopsy samples are obtained using a punch or shave biopsy method, but other methods can also be used, e.g., needle biopsies or excisional biopsies. Preferably the skin sample comprises skin from a reaction site.

The skin sample undergoes a skin culture and preparatory method for downstream analysis. The skin culture method places a portion, e.g., half, of a skin biopsy sample per well, e.g., for 2-7 days, 2-5 days, 3-5 days, or 3-4 days with a drug of interest (e.g., a potential allergic reaction trigger) and a second portion, e.g., the other half, biopsy with vehicle control. Other test compounds can be used in addition to known or approved drugs. The skin is then restimulated with the same drug or vehicle control and cultured for an additional 1-2 days. The skin is reviewed grossly for changes including loss of sample integrity or signs of cell death. Migratory T cells are collected, e.g., by pipetting the culture supernatant and cells non-adherent or loosely- adherent to skin biopsy into a collection tube, and then pelleting cells and removing the supernatant, prior to performing an enzymatic digestion with physical disaggregation to maximize the number of isolated viable T cells while minimizing structural cells, or alternatively keratinocyte/skin debris is collected intact (without enzymatic digestion and physical disaggregation). In addition, culture supernatant is collected..

In parallel, PBMCs are collected from the blood sample, e.g., using a standard Ficoll gradient. PBMCs potentially undergo selection based on findings in the skin samples; for example, for an SJS/TEN case collect CD8 and optionally CD14+ cells are collected, or for a spongiotic CD4+ appearing MDE reaction, the methods can include depleting Treg. E.g., CD8+ cytotoxic appearing reactions: CD8 T cells and CD14 positive monocytes are selected for culture if sufficient cell number. CD4+ T cell / spongiotic reactions or insufficient cell number: PBMCs are depleted of CD25+ (Treg) cells then cultured. PBMCs are stimulated as skin in Step 3.

Analysis is then performed to determine 1, 2, or all three readouts including (1) T cell receptor (TCR) repertoire analysis for clonal expansion; (2) biomarker analysis; and (3) skin integrity analysis.

The primary readout is based on T cell receptor (TCR) repertoire analysis on DNA extracted from the collected immune/T cells and residual skin biopsy tissue if not digested/disaggregated to identify clonal expansion. Methods for TCR sequencing include next generation sequencing (NGS) and single-cell TCR sequencing methods such as TCRseq or HT-TCRP sequencing; see, e.g., Mitchell and Michels, J Life Sci (Westlake Village). 2020 Dec; 2(4): 38-58; Fu et al., Front Immunol. 2021 Nov 5: 12:777756; Fbhse et al., Eur J Immunol. 2011 Nov;41(ll):3101-13. The presence of clonal expansion can be identified by comparison of frequency of each TCR clone to a threshold level that represents a control level of frequency associated with clones that have not undergone expansion, and a frequency above that threshold level indicates the presence of clonal expansion; preferably internal controls are used. In some embodiments, clonal expansion is defined as three times the frequency of nonexpanding clones within each sample. In some embodiments, the presence of specific clones identified as high-risk is indicative of a positive result independent of frequency.

The secondary readout is based on analyte analysis of the supernatant for particular biomarkers or a combination of biomarkers. Biomarkers analyzed can include one, two, three, or all four of granulysin, interferon gamma (IFNg), interleukin 5 (IL5), IL 10, and/or IL13; optionally granulysin is analyzed alone or with one, two, or three of IFNg, IL5, IL10, and/or IL-13. Methods for analyzing these biomarkers are known in the art and can include . The tertiary readout is gross analysis of the residual skin, e.g., for loss of integrity or signs of cell death. This can be performed visually, looking for signs of the skin sample breaking down, or using an assay for keratinocyte cell death such as detection of lactate dehydrogenase (LDH).

In some embodiments, HLA sequencing can be performed, e.g., using residual DNA from the collected cells. The HLA and TCR sequencing results from patient samples (active and/or resolved) can also be compared to a database comprising information regarding TCR/HLA/drug reactions, as added confirmation.

In some embodiments, the TCR sequencing results from resolved disease can be compared to sequences from samples obtained from patients during active clinical disease, if available, as added confirmation.

The methods can also be used to reduce the risk of severe drug reactions. In these methods, skin samples and blood are obtained from a patient prior to administration of a high risk medication. The skin and blood are analyzed with the drug as above. A result that indicates a risk of a drug reaction would indicate that a recommendation should be made to avoid the drug, or consider giving the drug only if clinically necessary but monitoring the patient closely; a result that indicates a low or no likelihood of a drug reaction would indicate that a recommendation can be made to take the drug as normal.

Medications associated with a high risk of dtDHR can include abacavir; allopurinol; amoxicillin; ampicillin; azithromycin; barbiturates; carbamazepine; cenobamate; cephalosporins, fluoroquinolones, e.g., ciprofloxacin/levofloxacin/gemifloxacin;clozapine; cotrimoxazole; dapsone; diclofenac; ethosuximide; isoniazid; lamotrigine; nevirapine; non-steroidal anti-inflammatory drugs; oseltamivir; oxicams; oxcarbazepine, paracetamol/acetaminophen; penicillins; phenytoin; sulfasalazine; sulfonamides; tetracyclines, vancomycin; valproate; and zonisamide. Any drug or vaccine (either active compound or excipient) can cause a reaction. The methods can use the following Table A to determine which action to take.

TABLE A - Risk Assessment and Treatment Algorithm

EXAMPLES

The invention is further described in the following examples, which do not limit the scope of the invention described in the claims.

MATERIALS AND METHODS

The following materials and methods were used in the Examples described herein.

Human studies

Retrospective analysis was conducted on FFPE skin samples of SJS/TEN, DRESS, MDE, and FDE from adult and pediatric patients from Brigham and Women’s Hospital (BWH), Boston Children’s Hospital (BCH), Dana Farber Cancer Institute, and Massachusetts General Hospital. All cases were clinically diagnosed as dtDHR by board-certified dermatologists, had pathology consistent with the diagnosis read by board-certified dermatopathologists and were vetted by a second board-certified dermatologist with expertise in dtDHRs (SJD). Alternative pathology or clinical diagnoses, cases lacking sufficient clinical data to confirm diagnosiswere excluded to eliminate potential for misdiagnosis. A second patient cohort from BWH was obtained by searching pathology cases of morbilliform drug eruption in patients that were lymphopenic (< 1000 lymphocytes/ml) at the time of skin biopsy.

Patients with clinically confirmed SJS/TEN or MDE were prospectively enrolled at BWH or Dana Farber Cancer Institute . A 6 mm punch biopsy, 40 ml peripheral blood, and medical record data were collected on each patient. All prospective study patients provided written informed consent prior to enrollment.

Human skin discarded during plastic or dermatologic surgeries, and PBMCs from blood bank leukopacks served as healthy controls.

Approval for human studies was granted by the Partners Healthcare Institutional Review Board (IRB), DFCI IRB and BCH IRB. Skin staining and microscopy

FFPE skin sections 5-6 mm thick were baked, deparaffinized, and rehydrated. Hematoxylin and eosin (H&E) staining was carried out by standard technique. For immunofluorescence staining, sections underwent antigen retrieval at 96°C, blocking of non-specific protein binding, and staining with primary then appropriate secondary antibodies. Primary antibodies used: anti-rabbit CD3 (A0452, Dako; polyclonal), antimouse CD45RO-biotinylated (304202, Biolegend; UCHL1), anti -mouse CD45RA (158-4D3; Novus, NBP2-15193), anti-mouse CD8 (M7103, Dako; C8/144B), and antirat CLA (321302, Biolegend; HECA-452). Secondary antibodies used, from Biolegend: AF555 goat anti-mouse IgG (Poly4053; 405324), AF555 donkey anti-rabbit IgG (Poly4064; 406412), AF488 Streptavidin (405235); from Invitrogen: AF647 goat antimouse IgG (Polyclonal; A-21236), AF647 goat anti-rabbit IgG (Polyclonal; A21245) and AF488 goat anti-rat IgM (Polyclonal; A21212). Sections were then counterstained with DAPI then imaged using the Mantra™ Quantitative Pathology Workstation and analyzed using InFORM analysis software (Akoya Biosciences).

Nanostring transcriptional profiling

RNA was extracted from FFPE skin biopsy scrolls using the RNeasy Micro Kit (Qiagen). Total RNA quantity and quality was measured using the BioDrop™ DUO spectrophotometer (Thomas Scientific) and a subset of samples was further evaluated using fragment analysis (Agilent Bioanalyzer, RNA NanoChip). RNA was concentrated as needed (RNA Clean & Concentrator Kit, Zymo Research). Samples were analyzed using a 200 gene custom designed panel (Table 1) from NanoString Technologies on an nCounter Digital Analyzer.

After processing, quality was assessed using NanoStringQCPro-v. 1.18.0 (82) and custom R code. Proportion of fields of view successfully counted, binding density, noise threshold, expression of positive and negative control genes, and expression of endogenous and housekeeping genes were evaluated. Samples that were found to be suboptimal were flagged and removed from the analysis. Raw counts were normalized with the geometric mean using first positive controls, then using a subset of housekeeping genes selected with an expression above a limit of detection (defined as mean expression of the negative control genes plus two times the standard deviation), and a mean value higher than 200 (value selected empirically based on average housekeeping expression). Any gene expressed below this limit was filtered out. Differential expression at the gene level was performed with the DESeq2-v.1.26.0 R package (83). Adjusted p-values (p _a dj) were estimated using the false discovery rate to correct for multiple comparisons. Primary analysis compared each dtDHR to healthy controls. Log2FC > ± 1, p _a dj < 0.05 was considered significant. Secondary analysis compared each form of dtDHR to each other. Log2FC > ± 1, p _a dj < 0.1 was considered significant.

All analyses were performed using R (v.3.6.0). Gene expression and volcano plots were made in Graphpad Prism (v9).

Prospective sample processing and staining

Samples were processed under sterile conditions. Upon collection, skin biopsies were halved and each half frozen in 500 ml Cryostor CS10 cry opreservation media (07930; Stemcell Technologies) in liquid nitrogen (LN2). PBMCs were collected from blood by Ficoll gradient and frozen in FBS + 10% DMSO in LN2. Skin and PBMC samples from three SJS/TEN patients, three MDE patients and six healthy control patients (three healthy control skin and three healthy control blood that were not paired) were processed for single-cell RNA-sequencing (scRNAseq) + Cellular Indexing of Transcriptomes and Epitopes sequencing (CITEseq) + T Cell Receptor seq (TCRseq) by the BWH Center for Cellular Profiling. Frozen skin biopsies were thawed for 30 secs in a 37°C water bath, rinsed in PBS, then thawed in 100% FBS on ice for 30 min. Biopsies were then rinsed in PBS, cut into small pieces and incubated with human whole skin dissociation kit without enzyme P (130-101-540; Miltenyi) for 2 hours 20 min at 37°C with agitation. Samples were washed in RPMI-1640 +10% FBS and centrifuged (1300 rpm, 4°C, 5 min). Skin was then disaggregated over a 70 pm filter, rinsed with RPMI-1640+10% FBS, and centrifuged (1300 rpm, 4°C, 5 min). PBMCs were thawed at 37°C and washed with RPMI-1640+10% FBS, then PBS. Cell pellets were resuspended in cold PBS. Cell counts and viability were determined by trypan blue exclusion.

Samples were stained with Zombie NIR Viability dye according to manufacturer’s instructions (423106; Biolegend). Cells were then washed with hash cell staining buffer (420201; Biolegend) and non-specific antibody binding was blocked by 5% Fc Receptor Blocking Solution (422301; Biolegend) for 10 min on ice. Cells were stained with: anti-human CD3, anti-human TotalSeq™-C hashing antibodies (Biolegend) as previously described (84), and TotalSeq™-C antibodies (Biolegend) (anti-human CD4, CD8, CD45RA, CD45RO, CD62L, CCR7, CD 127, CD 103, CD69, CD107a, FAS, CD56 and CD335) for 30 min on ice. After staining, cells were washed with cell staining buffer and resuspended in sorting buffer (IX PBS with 2.5mM EDTA, 25mM HEPES and 1% FBS) for flow sorting using a FACS Aria™ Fusion cell sorter (BD Biosciences). Live CD45 ⁺ CD3 ⁺ T cells were collected into PBS + 0.4% BSA.

Single cell 5’ mRNA sequencing

Single cell RNA-seq experiments were performed by the BWH Single Cell Genomics Core. Sorted CD3 ⁺ T cells from three skin and three blood samples were pooled and resuspended in 0.4% BSA in PBS at a concentration of 1,000 cells/pl, then loaded onto a single lane (Chromium chip A, 10X Genomics) followed by encapsulation in a lipid droplet (Single Cell 5 'kit VI, 10X Genomics), then by cDNA and library generation according to the manufacturer’s protocol. Three runs were performed for a total of 18 specimens.

5’ mRNA library was sequenced to an average of 50,000 reads per cell, V(D)J library and HTO (Cell Hashing antibodies) library sequenced to an average of 5,000 reads per cell, using Illumina Novaseq. 10X Genomics reads were processed with Cell Ranger v3.1 for gene expression (using GRCh38 as reference genome), CITEseq and hashtag oligo counts for demultiplexing and quantification of transcript counts per putative cell. Demultiplexing of pooled samples were performed using the Seurat R package (version 4.0, Satija Lab) (85) along with R v4.0.5 (86). Raw HTO counts were normalized with centered log ratio (CLR) transformation (NormalizeData func.), and then HTODemux function was used to demultiplex pooled samples and to filter out (i) negative cells with low HTO UMIs (empty droplets or failed reactions) and (ii) multiplet cells with high HTO UMIs (cells from multiple samples in a droplet) (84). Each of the three pooled samples were repeatedly demultiplexed and 18 samples were separated into distinct Seurat objects. We later pre-processed TCRseq reads with Cell Ranger v6.0.1 due to updates in the clonotype calling algorithm. We used the djvdj package (github.com/rnabioco/djvdj) to import processed TCRseq reads into R.

Pseudo-Bulk DE analysis

AggregateExpression function in Seurat package was used to aggregate RNA counts from each of 18 samples. DESeq and IfcShrink functions from DESeq2 package (87) were used to conduct the Pseudo-bulk analysis on aggregated counts. scRNAseq data analysis

Downstream analysis of the pre-processed scRNA data from 18 samples was performed using the Seurat R package. To elucidate the T cell heterogeneity within the blood and skin samples, whose transcripts and surface protein were jointly assayed, we implemented a combination of standard integration and weighted nearest neighbor analysis workflows of Seurat package (satijalab.org/seurat/articles/get_started.html).

Cells with (i) less than 1000 or more than 20000 total RNA UMI count, (ii) with more than 10000 ADT UMI counts, or (iii) whose total mitochondrial (MT) gene expression counts exceeded 20% of their total UMI counts were filtered out. RNA and ADT assays from all 18 samples were log normalized and centered log ratio (CLR) normalized, respectively. CD4 ⁺ CD8 ⁺ doublets and double positives were filtered out by removing cells having both more than 0.75 and 1 CLR normalized CD4 and CD8 ADT UMI counts, respectively. One healthy control skin sample was removed from the analysis given Seurat integration workflow failing due to low number of cells. This filter step resulted in a scRNA dataset of 15,084 cells.

The 2,000 most highly expressed genes in each of the 17 samples were detected using variance stabilizing transformation (vst) method (FindVariableFeatures func.). Ribosomal gene signatures were excluded from downstream analysis. All 8 skin and 9 blood samples were integrated with FindlntegrationAnchors and IntegrateData functions for RNA and ADT assays, using the first 12 and 30 dimensions of canonical correlation analysis, respectively. This step of batch correction across multiple samples resulted in two separate batch-corrected assays for RNA and ADT. We first scaled (ScaleData func.) and then dimensionally reduced (RunPCA func.) both integrated RNA and ADT assays using the first 12 and 30 principal components before constructing the multimodal (combined RNA and ADT) shared nearest neighbor graph with the FindMultiModalNeighbors function (48). We used the resulting weighted shared nearest neighbor (WSNN) graph to cluster the integrated dataset and to build a two-dimensional UMAP reduction of the dataset for further visual analysis (RunUMAP func.).

To detect the cellular heterogeneity of the integrated blood and skin datasets, we initially used FindCluster function with resolution parameters ranging from 0.4 to 1.8, using graph-based smart local moving algorithm (88). We conducted marker analysis using the Find AllMarkers function (with a log2FC threshold of 0.25 and for genes that are detected in more than 10% of all cells) to test for significantly highly expressed genes and surface proteins in each cluster against all other clusters, with Wilcoxon Rank Sum test. Separately conducted gene and protein marker analysis allowed us to choose resolution=0.6 which resulted in 14 distinct clusters. A cluster was removed for having cells with high percentage of MT counts and another cluster was also removed due to having only 2 cells which resulted in a final scRNA dataset of 14,681 cells. We isolated and reanalyzed some of these clusters to locate existing subclusters. For each subclustering analysis, we created separate Seurat objects given cells in each cluster, then rescaled and repeated dimensional reduction on both integrated RNA and ADT assays of the associated subset of cells. For each performed subclustering analysis, we re-executed FindMultiModalNeighbors function to build the WSNN and found existing subclusters using FindClusters function with resolutions ranging from 0.4 to 0.8 which identified 15 subclusters. This workflow generated 22 clusters in total (7 initial clusters and 15 subclusters).

Comparison of TCR sequence specificity to known sequences

We cross-referenced the TCR sequences of all expanded clones from the prospective study dtDHR patients to two publicly available databases, VDJdb (Bagaev DV et al, 2019) and McPAS-TCR (89), to identify potential antigenic specificity or cross-reactivity (Glanville et al., 2017; Huang H et al., 2020; Cottrell T et al., 2021). Our search included T cells with matching alpha and beta chains, or only alpha chains or only beta chains. We included the known HLA-A and B alleles of the prospective study patients when available in the databases. If not available, we searched against all HLA alleles available in the database. We also compared to published TCR sequences of drug-reactive clones (17, 54-56).

Mixed lymphocyte reaction and DNA extraction

Previously frozen PBMCs from SJS/TEN patient 1 were thawed at 37°C and washed with complete human media containing RPMI-1640 + 4% human AB serum (Sigma), 1% Penicillin/ Streptomycin, 2mM L-glutamine, O. lmM MEM non-essential amino acids, lOmM HEPES, ImM sodium pyruvate and 50pM beta-mercaptoethanol. Cell pellets were resuspended in complete human media. Cell counts and viability were determined by trypan blue exclusion. PBMCs were depleted of CD25+ cells using the EasySep™ Human Pan-CD25 Positive Selection and Depletion Kit (Stemcell technologies) to remove any potential Treg. CD25- T cells were recounted and plated in a 96-well flat bottom plate in triplicate in complete human media. Cells were incubated with culprit drug Bupropion HCL at concentrations of 50ng/ml and lOOng/ml, and water for injection (GIBCO) as a vehicle control. As a positive control, wells were precoated with 500ng/ml anti-human CD3 (Biolegend) for 2 hours at 37°C, followed by 3 washes with PBS. Cells were plated with 5pg/ml anti-human CD28. The 96-well plate was incubated at 37°C 5% CO2. Cells were restimulated with drug or vehicle control at day 4. On day 5, cells were collected and spun down with PBS. DNA was extracted using DNeasy blood and tissue kit per manufacturer’s instructions (Qiagen). The quantity and quality of DNA was checked using NanoDrop™ and DNA was sent for high-throughput TCRb gene sequencing using the immunoSEQ® Assay (Adaptive Biotechnologies, Seattle, USA).

Alternatively, skin biopsy was halved and cultured with Palbociclib 0.1 ng/ml or vehicle control for 4 days then re-stimulated with Palbociclib 0.1 ng/ml or vehicle for another day. Non-adherent T cells were collected. Skin biopsy was enzymatically digested as above with human whole skin dissociation kit (Miltenyi) and then disaggregated over a 70 pm strainer. Resultant single cells were combined with nonadherent T cells. PBMC were similarly cultured with Palbociclib at 3 concentrations 0.1, 1 and 10), with Binimetinib at 0.1 and 1 or with vehicle control for the same time period and restimulated as was skin. DNA was extracted from cultured skin and blood using the DNease blood and tissue kit per manufacturer’s instructions. Finally, DNA was extracted from FFPE skin biopsies (FDE or SJS/TEN) using DNeasy FFPE kit per manufacturer’s instructions (Qiagen). The quantity and quality of all DNA was checked using NanoDrop™ and DNA was sent for high-throughput TCRb gene sequencing using the immunoSEQ® Assay (Adaptive Biotechnologies, Seattle, USA).

Culture supernatant was assayed for granulysin by ELISA according to manufacturer’s instructions (R&D Systems)

Mice

HLA-B*57:01 C57B1/6J transgenic mice were provided by D. Margulies (National Institute of Immunology, Allergy and Infectious Disease) and M. Norcross (U.S. Food and Drug Administration) ⁶⁰ . Heterozygous mice were bred inhouse by crossing with C57BL/6J mice (Jackson Laboratories). Male and female mice between 6-12 weeks of age were used in experimentation. Mice were phenotyped to confirm transgene expression (60). Experiments were performed in accordance with the guidelines put forth by the Center for Animal Resources and Comparative Medicine at Harvard Medical School (HMS) (HMS IACUC approval #2016N000070).

HLA-B*57:01 positive ^(pos) mice or HLA-B*57:01 negative ^(neg) littermate controls were treated by intraperitoneal (i.p.) injection five days/week for 17 days with 3 mg of Abacavir (ABC) (188062-50-2; Sigma Aldrich) diluted in Water for Injection (WFI) ((60)). A subset of mice underwent concurrently treatment of the left ear topically via painting three days/week with 0.2 mg ABC in 30% ethanol. Dosing was based on the animal equivalent dosage (AED) of ABC (90). Vehicle control mice were treated with equal volume of WFI i.p. injection and painted topically with 30% ethanol when appropriate. All mice were depleted of CD4 ⁺ T cells by i.p. injection with 0.25 mg anti-CD4 mAb (clone GK1.5, BioXcell) in sterile PBS 3 days prior to and again on days 1, 4 and 7 during drug exposure. For in vivo challenge, mice were depleted of CD4 ⁺ T cells and treated systemically by i.p. injection with ABC or vehicle. Some mice were also treated with FTY720 (1 mg per/kg i.p. daily) (SML0700; Sigma Aldrich).

Mice were monitored thrice weekly for clinical signs of dermatitis and ear thickness measured by electronic digital caliper for the entirety of each study. Mice were harvested at peak of disease (day 22), at disease resolution (day 90), and at peak of drug challenge (day 107).

Mouse tissue harvest and processing

Blood was collected in heparin then PBMCs isolated by Ficoll gradient. A portion of ear was fixed in 10% neutral buffered formalin, embedded in paraffin and sectioned at 5 microns for H&E staining by standard laboratory method. Spleen, cervical lymph nodes, and remaining ear tissue were harvested into cold PBS. Spleens and lymph nodes were disaggregated over 70 pm strainers into single cell suspensions. Ear skin was cut into small pieces then incubated in 3 ml Hanks balanced salt solution (14175095; Thermofisher Scientific) supplemented with 1 mg ml ^-1 collagenase A (11088785103; Roche) and 40 pg ml’ ¹ DNase I (10104159001; Roche) at 37°C for 3 hours in a shaking incubator. To neutralize the collagenase, RPMI+10%FBS was added to the tubes and suspension spun down at 1400 rpm for 5 min at 4°C. The cell pellet was resuspended in complete RPMI media then disaggregated through a 70 pm filter. Cell counts and viability were assessed with trypan blue.

Mouse flow cytometry

Single cell suspensions of ear skin, blood, cervical lymph nodes and spleen were blocked for non-specific antibody binding with 5% Normal Goat Serum (NGS). All samples underwent viability staining with Zombie NIR viability dye (423106, Biolegend). Samples were then surface stained with combinations of: PE/Cy7-CD3 (17A2), PERCP-CD8a (53-6.7), APC/Cy7-CD8a (53-6.7), FITC-CD44 (IM7), PE- CD69 (H1.2F3), and APC-CD62L (MEL-14). CLA expression was assessed by incubating cells with E-Selectin/Fc Chimera (575-ES; R&D System) in conjunction with PerCP-conjugated F(ab’)2 fragments of goat anti-human IgG F(c) antibody (109- 126-170; Jackson ImmunoResearch). For intracellular cytokine staining cells were fixed and permeabilized using Cytofix/Cytoperm Fixation/Permeablization Kit (554714; BD Biosciences) according to manufacturer’s instructions and stained with FITC-IFN-y (XMG1.2), PE-TNF-a (MP6-XT22), and/or APC-Granzyme B (QA16A02) (Biolegend).

Statistical analysis

Two groups were compared using a two-tailed Mann-Whitney test in Graph Pad Prism (v9). When three groups were compared, a non-parametric Kruskal-Wallis test was performed, and if significant, was followed by Dunn’s test for multiple comparisons. P<0.05 was considered statistically significant.

Example 1. Retrospective clinical skin sample analysis

To begin to investigate T cell contribution to dtDHR, we analyzed FFPE skin samples previously collected for clinical purposes from patients with SJS/TEN, DRESS and MDE. Histologic analysis demonstrated typical findings associated with dtDHR (FIG. 1 A) (28, 29). SJS/TEN was marked by full thickness epidermal necrosis with pauci -inflammatory infiltrate, DRESS demonstrated a robust mononuclear infiltrate, while the infiltrate and reaction pattern observed in MDE was more variable though generally less dense. Immunofluorescence staining and microscopy confirmed the presence of skin homing (CLA ⁺ ), CD8 ⁺ and CD4 ⁺ (CD8‘) CD3 ⁺ T cell subsets (FIG. IB) in all forms of dtHDR.

Next, bulk transcriptional profiling using a custom-designed 200 gene panel from NanoString (Table 1) was performed on SJS/TEN, DRESS, MDE and healthy control FFPE skin samples. These samples were fixed immediately upon biopsy so results reflect transcription levels in situ, i.e. without any experimental manipulation. Given limited prior transcriptional analysis of all three forms of disease, our primary analysis compared each form of dtDHR to healthy skin. Differential gene expression analysis (DGEA) demonstrated that both severe forms of disease had significantly upregulated transcription of CD3E, CD8A, CD45RO, SELL (CD62L) and CCR7 compared to control skin, suggesting potential recruitment of central memory T cells (TCM) (30, 31) from secondary lymphoid organs (FIG. 1C). Comparatively, MDE lacked significant upregulation of these markers (FIG. 1C), suggesting that T cells were neither heavily recruited into nor extensively proliferating within MDE skin consistent with microscopic analysis. None of the three diseases demonstrated increased CD69 and ITGAE, skin TRM markers (32-34), or KLRG1, a terminally differentiated T cell (TEMRA) marker (35, 36) (Table 1). However, TRM may have low proliferative potential (32, 37-39) so increased gene expression would not necessarily be expected even if TRM were activated.

DGEA suggested that all three forms of disease were Thl/Tcl skewed. Genes for cytolytic granule components, GZMA, GZMB, and PRF1, and IFNg signature genes, CXCL9, CXCL10 and CXCL11 were significantly upregulated in mild and severe dtDHR compared to healthy control skin (FIGs. 1D-E). Analysis further demonstrated significantly increased transcription of GNLY in both SJS/TEN and DRESS, and of TNF in SJS/TEN alone compared to healthy controls (FIGs. 1D-E), similarly to prior reports (40-43).

Example 2. Prospective clinical sample analysis - scRNAseq + CITEseq

To more deeply interrogate T cell subsets during active disease, we prospectively studied T cells from skin and blood using scRNAseq + cellular indexing of transcriptomes and epitopes (CITE) seq + T cell receptor (TCR) seq of three SJS/TEN patients, three MDE patients, and three healthy controls. We limited the prospective study to SJS/TEN and MDE (and healthy controls) given the clearer divergence in immunologic milieu observed by histology and bulk transcriptional profiling between these two forms of dtDHR coupled with the technical and analytical complexity of prospective patient sample tissue and blood analysis by scRNAseq + CITEseq + TCRseq. Notably, MDE patient 2 had a robust skin reaction at risk for progression to severe disease with systemic involvement so received high-dose systemic steroids with clinical improvement (did not progress), while MDE patients 1 and 3 had mild reactions (Table 2).

Transcript and protein from all 17 samples were integrated using Weighted Nearest Neighbors algorithm (48), which resulted in 22 distinct cell clusters, with clear separation between CD4 ⁺ and CD8 ⁺ T cells (FIG. 2A; Table 3). Importantly, the addition of CITE-seq resulted in markedly improved resolution of 7 key phenotypic markers (CD45RA, CD45RO, CD62L, IL7Ra, CD69, CD 103 and CD56) compared to scRNAseq alone, consequently greatly improving cluster definition (FIG. 2 A, Table 3). Though CD4 and CD8 protein largely mirrored CD4 and CD8 mRNA, respectively, use of protein provided greater clarity than mRNA when defining clusters (FIG. 2A). The heat map in FIG. 2 A and Table 3 demonstrate the protein and mRNA used in defining each cluster.

The total number and percentage of T cells in each cluster in each patient is shown in Tables 4 and 5 and fold change in percentage of T cells in each cluster in each patient shown in Tables 6 and 7. Despite clear trend toward higher percentage of total CD8 ⁺ T cells in SJS/TEN compared to both MDE and healthy control skin, results were not statistically significant. Regarding differences between groups on an individual cluster level, only Treg 2 cluster was statistically significant, with markedly reduced Treg 2 in SJS/TEN skin compared to MDE skin (Kruskal -Wallis test with Dunn’s post-test, p=0.0373).

The percentage cytotoxic T cells of total T cells in skin correlated with more severe disease (FIGs. 2B). There was a statistically significant difference between the percentage of cytotoxic CD8 ⁺ T cells between SJS/TEN versus MDE skin. The increased percentage of cytotoxic CD8 ⁺ T cells in SJS/TEN coupled with the decrease in Treg resulted in skewing of cytotoxic T cell Treg ratio in SJS/TEN compared to MDE and healthy skin and blood (FIG. 2B). Cytotoxic T cells in skin in both SJS/TEN and MDE were distributed across multiple clusters, including CD8 ⁺ T effectors, CD8 ⁺ CD56 ⁺ T cells, CD8 ⁺ TEM, CD8 ⁺ TEMRA, and CD4 ⁺ T effectors, all classically considered recruited populations, as well as CD8 ⁺ skin TRM, and gd T cells which can be recruited or skin resident (49) (FIG. 2C), suggesting that multiple T cell subsets, both resident and recruited, contributed to disease. Results were largely paralleled in blood (FIG. 2C), though there was much lower percentage of cytotoxic T cells in blood in all clusters in MDE compared to SJS/TEN, supporting that SJS/TEN had greater systemic activation and recruitment of cytotoxic T cells through circulation into skin.

Pseudo-bulk DGEA (FIG. 3) confirmed a prominent CD8 ⁺ Thl/Tcl signature in SJS/TEN versus healthy skin, while CD4 and CCR8, a Treg recruiting chemokine (50), were decreased. The prominent cytotoxic skew was maintained when comparing SJS/TEN skin to MDE skin. Functional polarization appeared more variable in MDE skin than previously suggested by bulk transcriptional profiling, with markers of Tel (GZMB), Th2 (IL5, IL10 and I 13) and Treg (CTLA4, IL2RA and IL10) observed when compared to healthy control skin, potentially reflecting variability across MDE patients. Pseudo-bulk DGEA findings in blood complemented pseudo-bulk DGEA in skin and bulk transcriptional profiling in skin. SJS/TEN versus healthy control blood demonstrated a Thl skew (IFNG, TNF and CXCL9) as well as increased transcription of IL2, which promotes T cell expansion, and CCL20, CCR4 and CCR6 which promote T cell recruitment (57, 52). Notably, CCR4 binds CCL2 (57) and CCR6 binds CCL20 (55); both ligands were transcribed at increased levels in SJS/TEN skin by bulk transcriptional analysis Table 1) supporting recruitment of T cells from circulation into skin in SJS/TEN. Neither of these chemokine receptors were increased in T cells in MDE blood. Comparatively, MDE blood demonstrated significantly increased transcription of F0XP3, CTLA4, IL2RA and CCR8. CCR8 binds CCL18, which was increased in MDE skin by bulk transcriptional profiling (Table 1), suggesting a possible mechanism by which Treg may be recruited into skin in MDE. The skew toward Thl/Tcl and away from Treg was maintained in comparison of SJS/TEN to MDE blood.

Example 3. Prospective clinical sample analysis - TCRseq

Clonality analysis of every fully TCR sequenced T cell in each sample was performed. Results affirmed the above findings. The top 25 clones in skin and in blood, including their total number, their frequency, their amino acid sequence, whether they were detected in both skin and blood, and each patient’s corresponding HLA-A and HLA-B alleles are detailed in Table 8. The productive frequency (percentage of each clone with a fully sequenced TCR alpha and beta chain) of the top 25 clones in each sample was graphed, with clones detected in both skin and blood uniquely color coded while clones exclusive to one or the other tissue are shown in gray (FIG. 4A). Since there is no clear definition of an actively expanding clone by clonal frequency assessed in an individual sample, we began with a more stringent quantitative approach whereby clear clonal expansion was defined as 3x the frequency of non-expanding clones within each sample. Internal controls were used rather than comparison to healthy control samples, since healthy donors can have expansion from an active or recent antigenic exposure (for example a cold virus). Using this approach, SJS/TEN patients 1 and 3 and MDE patients 1, 2 and 3 had proliferation of one or more clones in skin, and SJS/TEN patients 1 and 2 and MDE patient 3 had expansion in blood (FIG. 4A). Given though that skin and blood were paired in this study, we next compared the top clones between skin and blood in each patient. The single expanded clone in SJS/TEN patient 1 skin and MDE patient 3 skin were each amongst the top expanded clones in blood. The top expanded clone in SJS/TEN patient 3 skin that had clearly expanded by clonal frequency, was also the top clone in that patient’s blood, suggesting that clonal expansion had occurred in blood, but the clonal frequency did not reach the threshold arbitrarily set by the above quantitative approach. Similarly, the top clone in skin of SJS/TEN patient 2 was the second highest clone in that patient’s blood, again raising the possibility of expansion in blood. Comparatively, the top expanded clone in skin of MDE patient 1 and 2 were not detected in blood suggesting potentially skin-limited expansion (FIG. 4A).

Given limitations of using clonal frequency alone to interpret active expansion, we capitalized on the scRNAseq + CITEseq data obtained concurrently to ascertain the phenotypes of the top expanded cells, with the presumption that actively expanding cells should fall predominantly into activated clusters. The top expanded clone in skin in each patient was distributed across multiple T cell phenotypic clusters including primarily activated clusters (FIG. 4B). In all three SJS/TEN patients, the top expanded clone in skin was a cytotoxic CD8 ⁺ T that spanned highly functional clusters in both skin and blood (FIG. 4B). Comparatively, the top expanded clones in MDE patients 1 and 2 skin were also cytotoxic CD8 ⁺ T cells that spanned highly functional clusters but were not detected in blood. Interestingly, the top expanded clone in MDE patient 3 skin and blood was a CD4 ⁺ T cell that spanned Treg clusters (FIG. 4B), mirroring the gene expression data from scRNAseq. Healthy control skin did not have expansion by the quantitative definition (FIG. 5A), and cluster analysis revealed that the top clone in skin of each healthy control skin sample was a CD4 ⁺ T cell spanning non-functional clusters (FIG. 5B). Taken together, the data support that clonal expansion occurred at least at low level in skin in all 6 patients, and that clonal expansion of a cytotoxic CD8 ⁺ T cell concurrently occurred in blood of all SJS/TEN patients but no MDE patient.

Presumably, the monoclonal or oligoclonal expansion and activation observed in these patient samples reflects drug-specific activation, given that these patients were all diagnosed with active dtDHR. It is theoretically possible that the expanded T cells reacted to antigen other than drug. In rare cases, SJS/TEN can occur secondary to infection, most commonly Mycoplasma or Herpes Simplex Virus. The clinical and histologic findings in MDE can similarly occur secondary to virus alone, or to the combination of drug plus virus (the classic example being rash induced when amoxicillin is given in EBV infection). None of the 6 prospective study patients had signs or symptoms of Mycoplasma, HSV, or other viral infection, supporting drug as the antigenic source in these patients. Further, comparison of our patients’ paired TCR alpha and beta chains and any unpaired alpha or beta chains to reported known TCR alpha or beta chains in the VDJdb and McPAS-TCR databases using our patients’ known HLA-A and HLA-B sequences when available in the database, and to all HLA- C, DR, DP, and DQ alleles reported in the database, revealed no overlapping sequences. The lack of overlap further argues against an infectious antigenic source, though importantly, these databases are by no means exhaustive of all infections or all HLA alleles.

Comparison of our patients’ TCR sequences to published sequences of expanded T cell clones in dtDHR revealed no shared sequences, though this is not surprising as there was no overlap between presumed culprit drug and HLA-A or HLA- B allele of our 6 patients with the published data (77, 54-56). Confirmation of drug specificity is historically challenging to demonstrate in the lab as drug reactivity assays (lymphocyte transformation tests) have historically poor sensitivity (57). Based on our clonality data we posited that clonal expansion could serve as a read-out for drug reactivity. Using residual PBMCs from SJS/TEN patient 1, we performed a mixed lymphocyte reaction in the presence or absence of presumed culprit drug and used high- throughput TCRP sequencing as a readout. Indeed, the top expanded clones in blood by TCRseq (FIG. 4A), expanded ex vivo in the presence of drug > 1.7 fold compared to vehicle control, while over 3000 clones did not expand (fold change < 1) in the presence of drug supporting that expansion was both drug and clone specific (FIG. 4C).

Example 4. Skin TRM can mediate MDE in the absence of circulating T cells

While the above data suggest that skin TRM may be a critical pathogenic T cell subset mediating MDE, there does appear to be at least some level of T cell recruitment from circulation into skin in some MDE cases. We therefore aimed to directly test the functional contribution of TRM to MDE, with the ideal model system containing only TRM. Anecdotally, MDE can develop in patients that are severely lymphopenic, intimating that skin TRM alone can mediate MDE. Importantly, though lymphopenic patients may have aberrant immune systems from their underlying disease or treatment that caused the lymphopenia, their MDE are essentially clinically and histopathologically indistinguishable to healthy patients. Lymphopenic patients also often react to the same common culprit drugs as healthy patients. This supports that immune features common to lymphopenic and non-lymphopenic patients are potentially causal in MDE, while disparate features are likely unnecessary for disease. We retrospectively identified 12 patients with clinically diagnosed MDE despite lymphopenia secondary to varying chemotherapy regimens for acute myelogenous leukemia (Table 9). We specifically chose patients with significant lymphopenia that was clearly secondary to their chemotherapy treatment since patients with dtDHR may develop lymphopenia as a result of their drug reactions. FFPE skin samples from 5 cases were available for analysis. H&E staining of lymphopenic specimens revealed a range in mononuclear cell infiltrate from mild to robust with both interface and spongiotic reaction patterns (FIG. 6A), similar to that observed in non-lymphopenic patients. Immunofluorescence staining confirmed a skin-homing CLA ⁺ CD3 ⁺ T cell infiltrate, consisting of both CD8 ⁺ and CD4 ⁺ (CD8‘) subsets that were predominantly CD45RO ⁺ (FIG. 6B). Despite the stark difference in the number of circulating lymphocytes (FIG. 6C), quantification showed similar numbers of CD3 ⁺ T cells per high powered field between lymphopenic and non-lymphopenic MDE and healthy controls (FIG. 6D). These data support that skin TRM surviving chemotherapy mediated the MDE despite the absence of circulating T cells.

Example 5. Skin TRM can mediate dtDHR in mice

Finally, we aimed to test the role of skin TRM versus recruited populations in disease in a mouse model in which all other variables are controlled (i.e., all mice had a healthy immune system). HLA-B*57:01 predisposes patients taking the drug abacavir to dtDHR (58, 59). Cardone et al. previously generated HLA-B*57:01Tg mice that developed CD8 ⁺ T cell-mediated ear dermatitis in response to topical plus systemic abacavir exposure coupled with CD4 ⁺ T cell depletion (60). In our hands, despite depletion of CD4 ⁺ T cells, treatment with systemic abacavir alone (i.e., in the absence of concurrent topical drug exposure) failed to induce a reaction (FIGs. 7A-B). Based on our human data, we hypothesized that this was because memory T cells, in particular skin TRM, are necessary to mediate disease, and naive mice lack a legitimate memory T cell pool, in particular skin TRM (67). We therefore modified the original mouse model to test our hypothesis. We treated HLA-B*57:01 ^pos mice or HLA-B*57:01 ^neg littermate controls depleted of CD4 ⁺ T cells with abacavir or vehicle systemically and topically to ear skin. Experimental mice developed a cytotoxic CD8 ⁺ T cell mediated skin-limited (MDE-like) reaction that was HLA-B*57:01 and drug dependent (FIG. 7C- D). In this setting, CD8 ⁺ T cells were primed in secondary lymphoid organs and migrated through blood into skin to mediate disease. Drug-induced dermatitis slowly resolved by day 90 (FIG. 7C). Ear thickness decreased but did not return to baseline as ears were scarred; however, active inflammation resolved based on clinical and histologic evaluation (FIG. 7D). Despite the absence of active inflammation, ear skin of HLA-B*57:01 ^pos , drug-treated mice demonstrated a CD8 ⁺ T cell population expressing CD62L ^low CD69 ⁺ CLA ⁺ consistent with skin TRM (FIGs. 7E-F). Concurrently, TEM (CD62L ^low CD44 ^high ) were observed in blood and TCM (CD62L ^high CD44 ^high ) in LN (FIG. 7E).

To confirm that drug-reactive memory T cells were generated by this method and investigate whether these memory T cells could mediate a true drug allergy, mice underwent in vivo drug challenge. At day 90, HLA-B*57:01 ^pos mice previously treated with abacavir or vehicle were now treated systemically with abacavir or vehicle by intraperitoneal injection without topical treatment (FIG. 8 A). Mice containing drug- reactive memory T cells developed an MDE-like reaction upon drug challenge, marked by increased ear thickness and clinically and histologically evident dermatitis, faster than the primary drug-exposed mice, consistent with a memory T cell response (FIGs. 8B-C). This reaction was drug specific as HLA-B*57:01 ^pos mice previously immunized against drug but now challenged with vehicle failed to develop a reaction. This reaction included expansion and migration of CD8 ⁺ T cells in lymph node and blood, with some migration into skin, indicating that this reaction was not purely a local immune response or percutaneous reaction (FIGs. 8D-E).

A subset of drug-challenged mice was concurrently treated with FTY720, an S1PR1 agonist that prevents egress of T cells from lymphoid organs (62, 63). These mice developed dermatitis only slightly delayed to non-FTY720 treated drug- challenged mice, who had the ability to recruit T cells to skin from secondary lymphoid organs (FIGs. 8B-C). Moreover, FTY720 treated mice had a slightly reduced number of CD8 ⁺ T cells yet equivalent or higher percentage and total number of functional CD8 ⁺ T cells in ear skin compared to non-FTY720 treated mice (FIGs. 8D-F), further supporting that the main mediators of disease were skin TRM. These data directly parallel our observations in lymphopenic patients that skin TRM appear to be sufficient to mediate skin-limited reactions.

Example 6. Clonal expansion occurs in fixed drug eruption.

Fixed drug eruptions (FDE) are another form of CD8 mediated dtDHR. They present with discrete lesions at one or more sites on the skin. TCR sequencing was performed on two patients who each had two biopsies taken simultaneously of FDE lesions (FIG. 9 A). Sequencing revealed a clear monoclonal or oligoclonal expansion at each site with the top clone shared between both sites. Comparatively, TCR sequencing was performed on a third patient who had skin biopsies taken from an active FDE lesion and a non-FDE lesion simultaneously (FIG. 9B). A monoclonal expansion was again observed in the FDE lesion, but not in the non-FDE skin, indicating drug- reactivity/specificity of that clone. Similar findings were observed in a fourth patient (FIG. 9C). Oligoclonal expansion was observed (two clones) in active FDE skin that were present only at low levels in non-FDE skin, again supporting drug-specificity of the clones. These two clones remained detectable in skin 6 years after disease resolution, supporting that skin could potentially be successfully assayed by clonal sequencing years after disease resolution.

Example 7. Clonal repertoire analysis can identify culprit drug in skin and blood in dtDHR patient.

A patient with SJS/TEN was prospectively enrolled. The patient provided a skin biopsy after treatment with systemic steroids and partial resolution of her rash. The skin biopsy was halved. One half was cultured with presumed culprit drug, palbociclib. The other half was cultured with vehicle control. After culture, T cells were isolated then DNA extracted and sequenced by high-throughput TCR sequencing. In addition, DNA was extracted from an FFPE skin biopsy taken at initial presentation of SJS/TEN, and another taken weeks later from a clinically separate rash. Clonal repertoire analysis revealed two drug-reactive clones, gray and purple, during initial presentation of SJS/TEN (FIG. 10 A) These clones expanded in culture with drug as demonstrated by increased frequency compared to vehicle control. The clones were present in skin weeks later after resolution of SJS/TEN, but were unexpanded during the unrelated rash, indicated drug/reaction specificity. The unrelated rash was mediated by a distinct clone (FIG. 10 A).

PBMCs were collected after disease resolution and cultured with vehicle, presumed culprit drug, palbociclib, or another drug that the patient had recently taken, binimetinib (FIG. 10B). One of the two drug-reactive clones in skin during active disease, the purple clone, was present in blood and expanded in culture with palbociclib at optimized concentration (FIG. 10B). Notably, the gray clone from skin was not detected in blood after disease resolution supporting that in some patients, assaying skin rather than blood may be critical. Finally, culture supernatants were collected and assayed by ELISA for granulysin. In both skin and PBMC assays, granulysin was increased in culture with palbociclib compared to vehicle control. Binimetinib did not increase granulysin levels (FIGs. 10C-D). These results confirm the clonality findings, that the patient reacted to/is allergic to palbociclib.

Table 1. Fold changes and adjusted p-values for dtDHR vs healthy skin by bulk transcriptional profiling SJS/TEN DRESS MDE SJS/TEN DRESS MDE SJS/TEN DRESS MDE

Significance = P _a dj <0.05 and Log2FC > ± 1 Table 2. Prospective study patient demographic and clinical data ^a Race/Ethnicity as patient-reported in the medical record ^b Immunosuppressive treatment (one or more doses) administered for dtDHR prior to study sample collection ^c Not available Table 3. Key markers used to define T cell clusters in scRNAseq + CITEseq + TCRseq.

Gene name is italicized if RNA was used; if protein was used, the protein name is not italicized.

Table 4. Number (and percentage) of T cells in skin per cluster per patient

5

Table 5. Number (and percentage) of T cells in blood per cluster per patient

Table 6. Log2FC of percentage of T cells in each cluster of each patient’s skin relative to healthy control skin n.f. non-functional

Table 7. Log2FC of percentage of T cells in each cluster of each patient’s blood relative to healthy control blood

Table 9. Clinical and histopathologic characteristics of lymphopenic AML patients with MDE a Chronic inflammation” on pathology refers to either lymphocytic or lymphohistocytic infiltrate bAbsolute lymphocyte count is reported from the day of biopsy, or day after biopsy if differential was not obtained on the day of biopsy. References

1. D. Y. Hsu, J. Brieva, N. B. Silverberg, J. I. Silverberg, Morbidity and Mortality of Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis in United States Adults. Journal of Investigative Dermatology 136, 1387-1397 (2016).

2. V. Velasco-Tirado, M. Alonso-Sardon, A. Cosano-Quero, A. Romero-Alegria, L. Sanchez-Los Arcos, A. Lopez-Bemus, J. Pardo-Lledias, M. Belhassen-Garcia, Lifethreatening dermatoses: Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis. Impact on the Spanish public health system (2010-2015). PLOS ONE 13, e0198582 (2018).

3. C. C. Lin, C. B. Chen, C. W. Wang, S. I. Hung, W. H. Chung, Stevens-Johnson syndrome and toxic epidermal necrolysis: risk factors, causality assessment and potential prevention strategies. Expert Rev Clin Immunol 16, 373-387 (2020).

4. L. M. Wheatley, M. Plaut, J. M. Schwaninger, A. Banerji, M. Castells, F. D. Finkelman, G. J. Gleich, E. Guttman-Yassky, S. A. Mallal, D. J. Naisbitt, D. A. Ostrov, E. J. Phillips, W. J. Pichler, T. A. Platts-Mills, J. C. Roujeau, L. B. Schwartz, L. A. Trepanier, Report from the National Institute of Allergy and Infectious Diseases workshop on drug allergy. J Allergy Clin Immunol 136, 262-271 e262 (2015).

5. S. G. Johansson, T. Bieber, R. Dahl, P. S. Friedmann, B. Q. Lanier, R. F. Lockey, C. Motala, J. A. Ortega Martell, T. A. Platts-Mills, J. Ring, F. Thien, P. Van Cauwenberge, H. C. Williams, Revised nomenclature for allergy for global use: Report of the Nomenclature Review Committee of the World Allergy Organization, October 2003. J Allergy Clin Immunol 113, 832-836 (2004).

6. P. G. H. a. C. Gell, R.R.A., in Clinical Aspects of Immunology, R. R. A. a. G. Coombs, P.G.H. , Ed. (Blackwell Science, 1963).

7. W.-H. Chung, S.-I. Hung, J.-Y. Yang, S.-C. Su, S.-P. Huang, C.-Y. Wei, S.-W. Chin, C.-C. Chiou, S.-C. Chu, H.-C. Ho, C.-H. Yang, C.-F. Lu, J.-Y. Wu, Y.-D. Liao, Y.-T. Chen, Granulysin is a key mediator for disseminated keratinocyte death in Stevens-Johnson syndrome and toxic epidermal necrolysis. Nature Medicine 14, 1343- 1350 (2008).

8. S. J. Posadas, A. Padial, M. J. Torres, C. Mayorga, L. Leyva, E. Sanchez, J. Alvarez, A. Romano, C. Juarez, M. Blanca, Delayed reactions to drugs show levels of perforin, granzyme B, and Fas-L to be related to disease severity. J Allergy Clin Immunol 109, 155-161 (2002). 9. W.-H. Chung, S.-I. Hung, H.-S. Hong, M.-S. Hsih, L.-C. Yang, H.-C. Ho, J.-Y. Wu, Y.-T. Chen, A marker for Stevens-Johnson syndrome. Nature 428, 486-486 (2004).

10. C.-Y. Cheng, S.-C. Su, C.-H. Chen, W.-L. Chen, S.-T. Deng, W.-H. Chung, HLA Associations and Clinical Implications in T-Cell Mediated Drug Hypersensitivity Reactions: An Updated Review. Journal of Immunology Research 2014, 1-8 (2014).

11. S. Mallal, E. Phillips, G. Carosi, J.-M. Molina, C. Workman, J. Tomazic, E. Jagel-Guedes, S. Rugina, O. Kozyrev, J. F. Cid, P. Hay, D. Nolan, S. Hughes, A. Hughes, S. Ryan, N. Fitch, D. Thorborn, A. Benbow, HLA-B*5701 Screening for Hypersensitivity to Abacavir. New England Journal of Medicine 358, 568-579 (2008).

12. C. Sukasem, T. Jantararoungtong, P. Kuntawong, A. Puangpetch, N. Koomdee, P. Satapompong, P. Supapsophon, J. Klaewsongkram, T. Rerkpattanapipat, HLA-B (*) 58:01 for Allopurinol-Induced Cutaneous Adverse Drug Reactions: Implication for Clinical Interpretation in Thailand. Front Pharmacol 7, 186 (2016).

13. B. Sousa-Pinto, C. Correia, L. Gomes, S. Gil-Mata, L. Araujo, O. Correia, L. Delgado, HLA and Delayed Drug-Induced Hypersensitivity. International Archives of Allergy and Immunology 170, 163-179 (2016).

14. T. Bellon, Mechanisms of Severe Cutaneous Adverse Reactions: Recent Advances. Drug Safety 42, 973-992 (2019).

15. O. Correia, L. Delgado, J. P. Ramos, C. Resende, J. A. Torrinha, Cutaneous T- cell recruitment in toxic epidermal necrolysis. Further evidence of CD8+ lymphocyte involvement. Arch Dermatol 129, 466-468 (1993).

16. L. Le Cleach, S. Delaire, L. Boumsell, M. Bagot, I. Bourgault-Villada, A. Bensussan, J. C. Roujeau, Blister fluid T lymphocytes during toxic epidermal necrolysis are functional cytotoxic cells which express human natural killer (NK) inhibitory receptors. Clinical and Experimental Immunology 119, 225-230 (2002).

17. A. P. Villani, A. Rozieres, B. Bensaid, K. K. Eriksson, A. Mosnier, F. Albert, V. Mutez, O. Brassard, T. Baysal, M. Tardieu, O. Allatif, F. Fusil, T. Andrieu, D. Jullien, V. Dubois, C. Giannoli, H. Gruffat, M. Pallardy, F.-L. Cosset, A. Nosbaum, O.

Kanagawa, J. L. Maryanski, D. Yerly, J.-F. Nicolas, M. Vocanson, Massive clonal expansion of polycytotoxic skin and blood CD8+ T cells in patients with toxic epidermal necrolysis. Science Advances 7, eabe0013 (2021). 18. A. Nassif, A. Bensussan, L. Boumsell, A. Deniaud, H. Moslehi, P. Wolkenstein, M. Bagot, J. C. Roujeau, Toxic epidermal necrolysis: effector cells are drug-specific cytotoxic T cells. J Allergy Clin Immunol 114, 1209-1215 (2004).

19. A. Nassif, A. Bensussan, G. Dorothee, F. Mami-Chouaib, N. Bachot, M. Bagot, L. Boumsell, J. C. Roujeau, Drug specific cytotoxic T-cells in the skin lesions of a patient with toxic epidermal necrolysis. J Invest Dermatol 118, 728-733 (2002).

20. I. Viard, P. Wehrli, R. Bullani, P. Schneider, N. Holler, D. Salomon, T. Hunziker, J. H. Saurat, J. Tschopp, L. E. French, Inhibition of toxic epidermal necrolysis by blockade of CD95 with human intravenous immunoglobulin. Science 282, 490-493 (1998).

21. M. Kinoshita, Y. Ogawa, N. Hama, I. Ujiie, A. Hasegawa, S. Nakajima, T. Nomura, J. Adachi, T. Sato, S. Koizumi, S. Shimada, Y. Fujita, H. Takahashi, Y. Mizukawa, T. Tomonaga, K. Nagao, R. Abe, T. Kawamura, Neutrophils initiate and exacerbate Stevens-Johnson syndrome and toxic epidermal necrolysis. Sci TranslMed 13, (2021).

22. R. A. Clark, Skin-resident T cells: the ups and downs of on site immunity. J Invest Dermatol 130, 362-370 (2010).

23. R. A. Clark, Resident memory T cells in human health and disease. Sci Transl Med 7, 269rv261 (2015).

24. J. M. Schenkel, D. Masopust, Tissue-resident memory T cells. Immunity 41, 886-897 (2014).

25. O. Gaide, R. O. Emerson, X. Jiang, N. Gulati, S. Nizza, C. Desmarais, H. Robins, J. G. Krueger, R. A. Clark, T. S. Kupper, Common clonal origin of central and resident memory T cells following skin immunization. Nat Med 21, 647-653 (2015).

26. S. J. Divito, A. T. Aasebo, T. R. Matos, P. C. Hsieh, M. Collin, C. P. Elco, J. T. O'Malley, E. S. Baekkevold, H. M. Reims, T. Gedde-Dahl, M. Hagerstrom, J. Hilaire, J. W. Lian, E. L. Milford, G. S. Pinkus, V. T. Ho, R. J. Soiffer, H. T. Kim, M. C. Mihm, Jr., J. Ritz, I. Guleria, C. S. Cutler, R. Clark, F. L. Jahnsen, T. S. Kupper, Peripheral host T cells survive hematopoietic stem cell transplantation and promote graft-versus- host-disease. J Clin Invest, (2020).

27. J. A. Trubiano, C. L. Gordon, C. Castellucci, S. N. Christo, S. L. Park, E. Mouhtouris, K. Konvinse, M. Rose, M. Goh, A. S. Boyd, E. J. Phillips, L. K. Mackay, Analysis of Skin-Resident Memory T Cells Following Drug Hypersensitivity Reactions. J Invest Dermatol, (2019). 28. B. Cote, J. Wechsler, S. Bastuji-Garin, H. Assier, J. Revuz, J. C. Roujeau, Clinicopathologic correlation in erythema multiforme and Stevens- Johnson syndrome. Arch Dermatol 131, 1268-1272 (1995).

29. E. M. Schunkert, S. J. Divito, Updates and Insights in the Diagnosis and Management of DRESS Syndrome. Curr Dermatol Rep 10, 192-204 (2021).

30. F. Sallusto, D. Lenig, R. Forster, M. Lipp, A. Lanzavecchia, Two subsets of memory T lymphocytes with distinct homing potentials and effector functions. Nature 401, 708-712 (1999).

31. F. Sallusto, J. Geginat, A. Lanzavecchia, Central Memory and Effector Memory T Cell Subsets: Function, Generation, and Maintenance. Annual Review of Immunology 22, 745-763 (2004).

32. R. Watanabe, A. Gehad, C. Yang, L. L. Scott, J. E. Teague, C. Schlapbach, C. P. Elco, V. Huang, T. R. Matos, T. S. Kupper, R. A. Clark, Human skin is protected by four functionally and phenotypically discrete populations of resident and recirculating memory T cells. Sci Transl Med 7, 279ra239 (2015).

33. L. K. Mackay, A. Braun, B. L. Macleod, N. Collins, C. Tebartz, S. Bedoui, F. R. Carbone, T. Gebhardt, Cutting edge: CD69 interference with sphingosine- 1 -phosphate receptor function regulates peripheral T cell retention. J Immunol 194, 2059-2063 (2015).

34. L. K. Mackay, A. Rahimpour, J. Z. Ma, N. Collins, A. T. Stock, M. L. Hafon, J. Vega-Ramos, P. Lauzurica, S. N. Mueller, T. Stefanovic, D. C. Tscharke, W. R. Heath, M. Inouye, F. R. Carbone, T. Gebhardt, The developmental pathway for CD103(+)CD8+ tissue-resident memory T cells of skin. Nat Immunol 14, 1294-1301 (2013).

35. D. Di Mitri, R. I. Azevedo, S. M. Henson, V. Libri, N. E. Riddell, R. Macaulay, D. Kipling, M. V. D. Soares, L. Battistini, A. N. Akbar, Reversible Senescence in Human CD4+CD45RA+CD27- Memory T Cells. The Journal of Immunology 187, 2093-2100 (2011).

36. S. Koch, A. Larbi, E. Derhovanessian, D. Ozcelik, E. Naumova, G. Pawelec, Multiparameter flow cytometric analysis of CD4 and CD8 T cell subsets in young and old people. Immun Ageing 5, 6 (2008).

37. J. C. Thome, Joseph, N. Yudanin, Y. Ohmura, M. Kubota, B. Grinshpun, T. Sathaliyawala, T. Kato, H. Lerner, Y. Shen, L. Farber, Donna, Spatial Map of Human T Cell Compartmentalization and Maintenance over Decades of Life. Cell 159, 814-828 (2014).

38. B. V. Kumar, W. Ma, M. Miron, T. Granot, R. S. Guyer, D. J. Carpenter, T. Senda, X. Sun, S. H. Ho, H. Lerner, A. L. Friedman, Y. Shen, D. L. Farber, Human Tissue-Resident Memory T Cells Are Defined by Core Transcriptional and Functional Signatures in Lymphoid and Mucosal Sites. Cell Rep 20, 2921-2934 (2017).

39. R. Bartolome-Casado, O. J. B. Landsverk, S. K. Chauhan, L. Richter, D. Phung,

V. Greiff, L. F. Risnes, Y. Yao, R. S. Neumann, S. Yaqub, O. 0yen, R. Homeland, E.

M. Aandahl, V. Paulsen, L. M. Sollid, S.-W. Qiao, E. S. Baekkevold, F. L. Jahnsen, Resident memory CD8 T cells persist for years in human small intestine. Journal of Experimental Medicine 216, 2412-2426 (2019).

40. W. H. Chung, S. I. Hung, J. Y. Yang, S. C. Su, S. P. Huang, C. Y. Wei, S. W. Chin, C. C. Chiou, S. C. Chu, H. C. Ho, C. H. Yang, C. F. Lu, J. Y. Wu, Y. D. Liao, Y. T. Chen, Granulysin is a key mediator for disseminated keratinocyte death in Stevens- Johnson syndrome and toxic epidermal necrolysis. Nat Med 14, 1343-1350 (2008).

41. M. Gandhi, A. Kowal-Vem, G. An, M. Hanumadass, Blister fluid composition in a pediatric patient with toxic epidermal necrolysis. J Burn Care Res 29, 671-675 (2008).

42. A. Nassif, H. Moslehi, S. Le Gouvello, M. Bagot, L. Lyonnet, L. Michel, L. Boumsell, A. Bensussan, J. C. Roujeau, Evaluation of the potential role of cytokines in toxic epidermal necrolysis. J Invest Dermatol 123, 850-855 (2004).

43. C. B. Chen, K. L. Kuo, C. W. Wang, C. W. Lu, R. Chung- Yee Hui, K. L. Lu,

W. C. Chang, W. T. Chen, F. Yun, Y. C. Teng, H. E. Lee, J. Y. Lin, H. C. Ho, M. H. Chi, Y. Yu-Wei Lin, C. J. Chang, Y. Lin, Jr., C. L. Ku, S. I. Hung, Y. C. Chang, W. H. Chung, C. Taiwan Severe Cutaneous Adverse Reaction, Detecting Lesional Granulysin Levels for Rapid Diagnosis of Cytotoxic T lymphocyte-Mediated Bullous Skin Disorders. J Allergy Clin Immunol Pract 9, 1327-1337 el323 (2021).

44. I. Azzaoui, S. A. Yahia, Y. Chang, H. Vorng, O. Morales, Y. Fan, N. Delhem, C. Pie, A. B. Tonnel, B. Wallaert, A. Tsicopoulos, CCL18 differentiates dendritic cells in tolerogenic cells able to prime regulatory T cells in healthy subjects. Blood 118, 3549-3558 (2011).

45. C. Chenivesse, Y. Chang, I. Azzaoui, S. Ait Yahia, O. Morales, C. Pie, A. Foussat, A. B. Tonnel, N. Delhem, H. Yssel, H. Vorng, B. Wallaert, A. Tsicopoulos, Pulmonary CCL18 recruits human regulatory T cells. J Immunol 189, 128-137 (2012). 46. W. A. Goodman, A. D. Levine, J. V. Massari, H. Sugiyama, T. S. McCormick, K. D. Cooper, IL-6 signaling in psoriasis prevents immune suppression by regulatory T cells. J Immunol 183, 3170-3176 (2009).

47. A. Schneider, S. A. Long, K. Cerosaletti, C. T. Ni, P. Samuels, M. Kita, J. H. Buckner, In active relapsing-remitting multiple sclerosis, effector T cell resistance to adaptive T(regs) involves IL-6-mediated signaling. Sci TranslMed 5, 170ral l5 (2013).

48. Y. Hao, S. Hao, E. Andersen-Nissen, W. M. Mauck, 3rd, S. Zheng, A. Butler, M. J. Lee, A. J. Wilk, C. Darby, M. Zager, P. Hoffman, M. Stoeckius, E. Papalexi, E. P. Mimitou, J. Jain, A. Srivastava, T. Stuart, L. M. Fleming, B. Yeung, A. J. Rogers, J. M. McElrath, C. A. Blish, R. Gottardo, P. Smibert, R. Satija, Integrated analysis of multimodal single-cell data. Cell 184, 3573-3587 e3529 (2021).

49. R. Castillo-Gonzalez, D. Cibrian, F. Sanchez-Madrid, Dissecting the complexity of gammadelta T-cell subsets in skin homeostasis, inflammation, and malignancy. J Allergy Clin Immunol 147, 2030-2042 (2021).

50. A. lellem, M. Mariani, R. Lang, H. Recalde, P. Panina-Bordignon, F. Sinigaglia, D. D' Ambrosio, Unique chemotactic response profile and specific expression of chemokine receptors CCR4 and CCR8 by CD4(+)CD25(+) regulatory T cells. J Exp Med 194, 847-853 (2001).

51. T. Zhang, R. Somasundaram, K. Berencsi, L. Caputo, P. Gimotty, P. Rani, D. Guerry, R. Swoboda, D. Herlyn, Migration of cytotoxic T lymphocytes toward melanoma cells in three-dimensional organotypic culture is dependent on CCL2 and CCR4. Eur J Immunol 36, 457-467 (2006).

52. T. Yamazaki, X. O. Yang, Y. Chung, A. Fukunaga, R. Nurieva, B. Pappu, N. Martin-Orozco, H. S. Kang, L. Ma, A. D. Panopoulos, S. Craig, S. S. Watowich, A. M. Jetten, Q. Tian, C. Dong, CCR6 regulates the migration of inflammatory and regulatory T cells. J Immunol 181, 8391-8401 (2008).

53. M. Baba, T. Imai, M. Nishimura, M. Kakizaki, S. Takagi, K. Hieshima, H. Nomiyama, O. Yoshie, Identification of CCR6, the specific receptor for a novel lymphocyte-directed CC chemokine LARC. The Journal of biological chemistry 272, 14893-14898 (1997).

54. W. H. Chung, R. Y. Pan, M. T. Chu, S. W. Chin, Y. L. Huang, W. C. Wang, J. Y. Chang, S. I. Hung, Oxypurinol-Specific T Cells Possess Preferential TCR Clonotypes and Express Granulysin in Allopurinol-Induced Severe Cutaneous Adverse Reactions. J Invest Dermatol 135, 2237-2248 (2015). 55. C. W. Yang, S. I. Hung, C. G. Juo, Y. P. Lin, W. H. Fang, I. H. Lu, S. T. Chen, Y. T. Chen, HLA-B*1502-bound peptides: implications for the pathogenesis of carbamazepine-induced Stevens-Johnson syndrome. J Allergy Clin Immunol 120, 870- 877 (2007).

56. R. Y. Pan, M. T. Chu, C. W. Wang, Y. S. Lee, F. Lemonnier, A. W. Michels, R. Schutte, D. A. Ostrov, C. B. Chen, E. J. Phillips, S. A. Mallal, M. Mockenhaupt, T. Bellon, W. Tassaneeyakul, K. D. White, J. C. Roujeau, W. H. Chung, S. I. Hung, Identification of drug-specific public TCR driving severe cutaneous adverse reactions. Nat Commun 10, 3569 (2019).

57. M. R. Ardern-Jones, M. Mockenhaupt, Making a diagnosis in severe cutaneous drug hypersensitivity reactions. Curr Opin Allergy Clin Immunol 19, 283-293 (2019).

58. S. Mallal, D. Nolan, C. Witt, G. Masel, A. M. Martin, C. Moore, D. Sayer, A. Castley, C. Mamotte, D. Maxwell, I. James, F. T. Christiansen, Association between presence of HLA-B*5701, HLA-DR7, and HLA-DQ3 and hypersensitivity to HIV-1 reverse-transcriptase inhibitor abacavir. Lancet 359, 727-732 (2002).

59. S. Mallal, E. Phillips, G. Carosi, J. M. Molina, C. Workman, J. Tomazic, E. Jagel-Guedes, S. Rugina, O. Kozyrev, J. F. Cid, P. Hay, D. Nolan, S. Hughes, A. Hughes, S. Ryan, N. Fitch, D. Thorborn, A. Benbow, P.-S. Team, HLA-B*5701 screening for hypersensitivity to abacavir. N Engl J Med 358, 568-579 (2008).

60. M. Cardone, K. Garcia, M. E. Tilahun, L. F. Boyd, S. Gebreyohannes, M. Yano, G. Roderiquez, A. D. Akue, L. Juengst, E. Mattson, S. Ananthula, K. Natarajan, M. Puig, D. H. Margulies, M. A. Norcross, A transgenic mouse model for HLA-B*57:01- linked abacavir drug tolerance and reactivity. Journal of Clinical Investigation 128, 2819-2832 (2018).

61. L. K. Beura, S. E. Hamilton, K. Bi, J. M. Schenkel, O. A. Odumade, K. A. Casey, E. A. Thompson, K. A. Fraser, P. C. Rosato, A. Filali-Mouhim, R. P. Sekaly, M. K. Jenkins, V. Vezys, W. N. Haining, S. C. Jameson, D. Masopust, Normalizing the environment recapitulates adult human immune traits in laboratory mice. Nature 532, 512-516 (2016).

62. Z. J. Luo, T. Tanaka, F. Kimura, M. Miyasaka, Analysis of the mode of action of a novel immunosuppressant FTY720 in mice. Immunopharmacology 41, 199-207 (1999). 63. V. Brinkmann, J. G. Cyster, T. Hla, FTY720: Sphingosine 1 -Phosphate Receptor- 1 in the Control of Lymphocyte Egress and Endothelial Barrier Function. American Journal of Transplantation 4, 1019-1025 (2004).

64. E. M. Schunkert, P. N. Shah, S. J. Divito, Skin Resident Memory T Cells May Play Critical Role in Delay ed-Type Drug Hypersensitivity Reactions. Front Immunol 12, 654190 (2021).

65. A. P. Villani, A. Rozieres, B. Bensaid, K. K. Eriksson, A. Mosnier, F. Albert, V. Mutez, O. Brassard, T. Baysal, M. Tardieu, O. Allatif, F. Fusil, T. Andrieu, D. Jullien, V. Dubois, C. Giannoli, H. Gruffat, M. Pallardy, F.-L. Cosset, A. Nosbaum, O.

Kanagawa, J. L. Maryanski, D. Yerly, J.-F. Nicolas, M. Vocanson, Massive clonal expansion of polycytotoxic skin and blood CD8+ T cells in patients with toxic epidermal necrolysis. Science Advances 7, eabe0013 (2021).

66. R. J. Lehloenya, T. Isaacs, T. Nyika, A. Dhana, L. Knight, S. Veenstra, J. Peter, Early high-dose intravenous corticosteroids rapidly arrest Stevens Johnson syndrome and drug reaction with eosinophilia and systemic symptoms recurrence on drug reexposure. J Allergy Clin Immunol Pract 9, 582-584 e581 (2021).

67. H. Azukizawa, Animal models of toxic epidermal necrolysis. The Journal of Dermatology 38, 255-260 (2011).

68. R. Takahashi, Y. Kano, Y. Yamazaki, M. Kimishima, Y. Mizukawa, T.

Shiohara, Defective regulatory T cells in patients with severe drug eruptions: timing of the dysfunction is associated with the pathological phenotype and outcome. J Immunol 182, 8071-8079 (2009).

69. N. Yoshioka, A. Suto, R. Abe, N. Saito, J. Murata, I. Hayashi-Ujiie, D. Hoshina, Y. Fujita, H. Shimizu, Disturbed balance in three subpopulations of CD4(+)Foxp3(+) regulatory T cells in Stevens- Johnson syndrome and toxic epidermal necrolysis patients. Clinical immunology 148, 89-91 (2013).

70. R. Pavlos, K. D. White, C. Wanjalla, S. A. Mallal, E. J. Phillips, Severe Delayed Drug Reactions: Role of Genetics and Viral Infections. Immunol Allergy Clin North Am 37, 785-815 (2017).

71. D. Picard, B. Janela, V. Descamps, M. D'lncan, P. Courville, S. Jacquot, S. Rogez, L. Mardivirin, H. Moins-Teisserenc, A. Toubert, J. Benichou, P. Joly, P. Musette, Drug reaction with eosinophilia and systemic symptoms (DRESS): a multi organ antiviral T cell response. Sci TranslMed 2, 46ra62 (2010). 72. C. A. Almeida, P. van Miert, K. O'Driscoll, Y. M. Zoet, A. Chopra, C. Witt, M. John, F. H. J. Claas, L. J. D'Orsogna, Virus-specific T-cell clonotypes might contribute to drug hypersensitivity reactions through heterologous immunity. J Allergy Clin Immunol 144, 608-611 e604 (2019).

73. B. V. Kumar, T. J. Connors, D. L. Farber, Human T Cell Development, Localization, and Function throughout Life. Immunity 48, 202-213 (2018).

74. M. D. Martin, V. P. Badovinac, Defining Memory CD8 T Cell. Front Immunol 9, 2692 (2018).

75. P. C. Rosato, S. Wijeyesinghe, J. M. Stolley, D. Masopust, Integrating resident memory into T cell differentiation models. Current opinion in immunology 63, 35-42 (2020).

76. M. M. Klicznik, P. A. Morawski, B. Hollbacher, S. R. Varkhande, S. J. Motley,

L. Kuri -Cervantes, E. Goodwin, M. D. Rosenblum, S. A. Long, G. Brachtl, T. Duhen,

M. R. Betts, D. J. Campbell, I. K. Gratz, Human CD4(+)CD103(+) cutaneous resident memory T cells are found in the circulation of healthy individuals. Sci Immunol 4, (2019).

77. R. Fonseca, L. K. Beura, C. F. Quamstrom, H. E. Ghoneim, Y. Fan, C. C. Zebley, M. C. Scott, N. J. Fares-Frederickson, S. Wijeyesinghe, E. A. Thompson, H. Borges da Silva, V. Vezys, B. Youngblood, D. Masopust, Developmental plasticity allows outside-in immune responses by resident memory T cells. Nat Immunol 21, 412- 421 (2020).

78. F. M. Behr, L. Parga- Vidal, N. A. M. Kragten, T. J. P. van Dam, T. H. Wesselink, B. S. Sheridan, R. Arens, R. A. W. van Lier, R. Stark, K. van Gisbergen, Tissue-resident memory CD8(+) T cells shape local and systemic secondary T cell responses. Nat Immunol 21, 1070-1081 (2020).

79. J. Strobl, L. M. Gail, L. Kleissl, R. V. Pandey, V. Smejkal, J. Huber, V. Puxkandl, L. Unterluggauer, R. Dingelmaier-Hovorka, D. Atzmuller, T. Krausgruber, C. Bock, P. Wohlfarth, W. Rabitsch, G. Stary, Human resident memory T cells exit the skin and mediate systemic Th2-driven inflammation. Journal of Experimental Medicine 218, (2021).

80. K. L. Cepek, S. K. Shaw, C. M. Parker, G. J. Russell, J. S. Morrow, D. L. Rimm, M. B. Brenner, Adhesion between epithelial cells and T lymphocytes mediated by E-cadherin and the alpha E beta 7 integrin. Nature 372, 190-193 (1994). 81. P. I. Karecla, S. J. Bowden, S. J. Green, P. J. Kilshaw, Recognition of E- cadherin on epithelial cells by the mucosal T cell integrin alpha M290 beta 7 (alpha E beta 7). Eur J Immunol 25, 852-856 (1995).

82. S. T. Nickles D, Ziman R and Bourgon R, NanoStringQCPro: Quality metrics and data processing methods for NanoString mRNA gene expression data. R package version 1.8.0, (2017).

83. M. I. Love, W. Huber, S. Anders, Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biology 15, (2014).

84. M. Stoeckius, S. Zheng, B. Houck-Loomis, S. Hao, B. Z. Yeung, W. M. Mauck, P. Smibert, R. Satija, Cell Hashing with barcoded antibodies enables multiplexing and doublet detection for single cell genomics. Genome Biol 19, 224 (2018).

85. T. Stuart, A. Butler, P. Hoffman, C. Hafemeister, E. Papalexi, W. M. Mauck, 3rd, Y. Hao, M. Stoeckius, P. Smibert, R. Satija, Comprehensive Integration of SingleCell Data. Cell 177, 1888-1902 el821 (2019).

86. . (2021).

87. A. Zhu, J. G. Ibrahim, M. I. Love, Heavy-tailed prior distributions for sequence count data: removing the noise and preserving large differences. Bioinformatics 35, 2084-2092 (2019).

88. v. E. N. Waltman L, A smart local moving algorithm for large-scale modularitybased community detection. Eur Phys J B 86, 471 (2013).

89. N. Tickotsky, T. Sagiv, J. Prilusky, E. Shifrut, N. Friedman, McPAS-TCR: a manually curated catalogue of pathology-associated T cell receptor sequences. Bioinformatics 33, 2924-2929 (2017).

90. A. Nair, S. Jacob, A simple practice guide for dose conversion between animals and human. Journal of Basic and Clinical Pharmacy 7, 27 (2016).

91. Watanabe, R. et al. Human skin is protected by four functionally and phenotypically discrete populations of resident and recirculating memory T cells. Sci Trans/ Med 7, 279ra239, doi: 10.1126/scitranslmed.3010302 (2015).

OTHER EMBODIMENTS

It is to be understood that while the invention has been described in conjunction with the detailed description thereof, the foregoing description is intended to illustrate and not limit the scope of the invention, which is defined by the scope of the appended claims. Other aspects, advantages, and modifications are within the scope of the following claims.