Single-cell immunopathology of systemic contact allergy associated with corticosteroids

Allergic contact dermatitis (CD) is a delayed hypersensitivity reaction with keratinocyte-directed cytotoxicity mediated by CD8⁺ T-cells (1). Despite association with topical exposure, in rare instances presentation may follow systemic exposure to corticosteroids following prior sensitisation through skin (2). Among corticosteroids, methylprednisolone and other Coopman group ‘A’ structures classified by structural substitution on C16 of the D ring are thought more immunogenic based on prevalence of reporting (3), but rarity of systemically-induced CD (SCD) has limited investigation of cellular immunopathogenesis. Comparing T-cells from methylprednisolone-positive intradermal test (IDT) to those of unaffected skin from a patient recovered from SCD, we utilised single-cell T-cell receptor (TCR)-RNA-sequencing to define (i) αβ TCR clonality and (ii) the transcriptome signature of T-cell effectors at the site of tissue damage.

The case patient was a 50-year-old female who developed SCD with local edema (Figure 1A, i) and delayed spreading rash (Figure 1A, ii) following interarticular injection of dexamethasone and methylprednisolone-acetate (MA) into her feet as treatment for a history of dry eyes and arthralgia, respectively. Three months post-recovery, IDT was performed on the lower left arm, unaffected during reaction, using structurally diverse corticosteroids with only MA producing a delayed positive reading (Figure 1B). At six months, follow-up IDT was performed with methylprednisolone-sodium-succinate (MSS), and prick testing to PEG3350, an excipient with reported allergic immunogenicity (4). A positive IDT observed with MSS at 8 hours became more indurated, documented at 24 and 48 hours, with no similar reaction observed to PEG3350 nor other corticosteroids (Figure 1C) including Triamcinolone where site was discoloured by minor bruising. Histopathology from IDT+ skin recorded a perivascular lymphocytic dermal infiltrate (Figure 1D).

Figure 1. Clinical reaction.

(A) Primary reaction (i) edema at injection site in feet, (ii) delayed spreading rash. (B) Initial IDT to a corticosteroid panel with saline/histamine controls at (i) 24 and (ii) 48 hours. (C) Follow-up IDT to methylprednisolone-sodium-succinate (MSS; 0.5-5 mg/mL) and PEG3350 at (i) 24 and (ii) >30 hours. (D) Hematoxylin and eosin stain on IDT skin.

With confirmed causality, 4 mm punch biopsies were taken from (i) site of positive 5 mg/mL MSS+ IDT and (ii) upper arm as unaffected control and processed via liberase-based digestion. Briefly, skin biopsies were incubated (90 mins) with 2.5 mg/mL Liberase and 20 mg/mL DNase I, and suspension filtered using a 70-μm cell strainer. Resulting filtrate was then washed thrice before flow staining using anti-human CD3 A700 (UCHT1). With inclusion of viability dye, live CD3+ T-cells were isolated as interest effectors using a BD FACSAria III flow sorter into SmartSeq buffer and stored at −80°C until sequencing. An 18-fold greater abundance of CD3+ T-cells were isolated from MSS+ IDT versus unaffected skin.

Using an altered method of SMARTseq2 for scTCR-RNA-sequencing, thawed sample in 96-well plates underwent oligo dT-primed reverse transcription. Full-length amplified cDNA libraries were generated using generic PCR handles and TCR gene transcripts amplified using TCR-specific primers. cDNA was fragmented with subsequent amplification of 3’ and 5’ ends in which barcoded amplicons were pooled and sequenced on an Illumina Nextseq 500. Data was demultiplexed, aligned to reference genome, and UMIs used to remove PCR-duplicate reads. Cells with <200 genes and >5% mitochondrial content were removed. Downstream analyses were performed in the Seurat v.2.3.4 R package and Visual Genomics Analysis Studio (5).

With heterodimeric pairing of functional αβ TCR, scTCR-seq identified a polyclonal repertoire in both unaffected and affected skin (Figure 2A) without immunodominance nor shared representation, indicative of immune turnover and influx of reactive T-cells to skin during reaction. While dominant TCR are reported in cutaneous blisters during carbamazepine-induced Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis (6), we and others previously report polyclonal response during abacavir hypersensitivity, including positive patch-test, aligning with response driven by an altered self-peptide repertoire (7). Further study is warranted to determine (i) the TCR-interacting moieties during MA-induced SCD and (ii) whether polyclonal response is representative of primary reaction.

Figure 2. scTCR-RNA-seq analyses of CD3+ T-cells from unaffected and IDT-positive biopsy.

(A) Clonality of TCRαβ pairings. Circos segments proportionate to dominance (increasing green to red). (B) Differential gene expression (one-way ANOVA). Genes with p ≤ 0.05 coloured to unaffected (blue) or affected (red) skin. (C) Box plot expression of (i) DEG interest genes alongside (ii) housekeeping gene β2M.

To ascertain the signature of cellular effectors we performed differential expression analyses between T-cells of unaffected (n = 147 cells) and affected (n= 231 cells) skin, finding 515 and 169 differentially expressed genes (DEG, p ≤0.05), respectively (Figure 2B). Pathway analysis of those in affected skin (www.genome.jp/kegg/) aligned with T-cell activation, with those in the top 10 including glycogen metabolism, MAPK, calcium, TLR, and chemokine signalling (data not shown). Subsequent analysis highlighted interest markers among the top DEG of T-cells from affected skin with P <0.001; ZYG11A implicated in cell division, PTMA an inducer of proliferation, GRAP2 regulating lymphocyte development, and ZHX1 for proliferation, migration, and invasion. Heat shock coordinator EEF1A1 was also more highly expressed implicating pathways associated with stress survival (P < 0.001). Box plots are shown for these markers ( Figure 2C(i)) alongside housekeeping gene β2M (Figure 2C(ii)), detailing subtle yet significant differences in expression.

We further sought to understand recruitment of T-cells to affected skin as implied by scTCR-seq. The defining tissue-resident memory (Trm) gene ITGAE was not differentially expressed (P = 0.210), with low expression across samples (Supplementary figure S1a). The majority of T-cells also lacked expression of sphingosine-1-phosphate receptor 1 (S1PR1), an inhibitor of differentiation towards T-cell residency (8), and CCR7 (Supplementary figure S1a), which facilitates efflux from peripheral tissues (9), aligning with recent influx of T-cells destined for at least short-term retention in tissue.

Given the activated effector signature, we investigated the regulatory pressures imparted. Expression of co-inhibitory receptors PDCD1 (P =0.426), CTLA4 (P =0.665), HAVCR2 (P =0.500) and LAG3 (P =0.262) were not highly or differentially expressed, nor T-reg markers IL2RA (P =0.104) and FOXP3 (P =0.434). Indeed, only one cell from total sample met minimum (>0.5) confidence for T-reg classification using a Seurat module score and Azimuth reference signature (azimuth.hubmapconsortium.org; data not shown). Interestingly, the co-inhibitory marker LGALS1 (P <0.001) that plays a role in skin healing (10) showed higher expression in unaffected skin (Supplementary figure S1b), indicating that loss of expression may have a role in differential skin involvement, warranting further investigation.

Our case demonstrates a history consistent with SCD showing selective IDT positive to methylprednisolone. Importantly from IDT+ biopsy, we isolated T-cells that displayed a uniquely polyclonal TCR repertoire and an activated, proliferative (GRAP2⁺, ZYG11A⁺, PTMA⁺, ZHX1⁺) and non-Trm (ITGAE⁻) signature indicative of an influx of effector T-cells. Thus, for the first time in the context of corticosteroid hypersensitivity, we demonstrate utility of IDT biopsy and single-cell TCR-RNA-sequencing for discovery of tissue-relevant effector biomarkers to enhance diagnosis of SCD.

Supplementary Material

Supplementary Figure S1

Figure S1. Box plot expression of investigational genes for (A) tissue-residency (ITGAE, S1PR1, CCR7) and (B) immune-regulation (PDCD1, CTLA4, HAVCR2, LAG3, FOXP3, IL2RA, LGALS1) between T-cells of unaffected (blue) and IDT-positive skin (red).

Data Avaliability Statement.

Upon publication, single-cell datasets in this article can be accessed at http://dx.doi.org/10.17632/6nvh5jrn74.1, hosted at Mendeley Data V1 (Gibson, Andrew (2021), “2021-1 Single-cell immunopathology of systemic contact allergy associated with corticosteroids.”, Mendeley Data, V1, doi: 10.17632/6nvh5jrn74.1)."