Abstract
Background:
Prurigo nodularis (PN) is a debilitating, difficult to treat, intensely pruritic, chronic inflammatory skin disease characterized by hyperkeratotic skin nodules. The pathogenesis of PN is not well understood but is believed to involve cross talk between sensory nerve fibers, immune cells, and the epidermis. It is centered around the neuroimmune cytokine IL-31, driving an intractable itch-scratch cycle.
Objective:
To provide a comprehensive view of the transcriptomic changes in PN skin and characterize the mechanism of action of the anti-IL-31 receptor inhibitor nemolizumab. Method: RNA-sequencing of biopsy samples obtained from a cohort of patients treated with the anti-IL-31 receptor inhibitor nemolizumab and taken at baseline and week 12. Generation and integration of patient data with RNA-seq data generated from reconstructed human epidermis stimulated with IL-31 and other pro-inflammatory cytokines.
Results:
Our results demonstrate that nemolizumab effectively decreases IL-31 responses in PN skin leading to effective suppression of downstream inflammatory responses including Th2/IL-13 and Th17/IL-17 responses. This is accompanied by decreased keratinocyte proliferation and normalization of epidermal differentiation and function. Furthermore, our results demonstrate how transcriptomic changes associated with nemolizumab treatment correlate with improvement in lesions, pruritus, stabilization of extracellular matrix remodeling, and processes associated with cutaneous nerve function.
Conclusion:
In summary, these data demonstrate a broad response to IL-31 receptor inhibition with nemolizumab and confirm the critical upstream role of IL-31 in PN pathogenesis.
Keywords: Prurigo nodularis, nemolizumab, Th2, Th17, mechanism
CAPSULE SUMMARY
Prurigo Nodularis pathophysiology is complex, involving both Th2 and Th17 responses. Treatment of patients with Prurigo Nodularis with the anti-IL-31RA agent nemolizumab results in decreased Th2/IL-13/IL-31 and Th17/IL-17 responses and normalization of epidermal proliferation.
INTRODUCTION
Prurigo nodularis (PN) is an intensely pruritic, chronic inflammatory skin disease characterized by hyperkeratotic skin nodules. While the pathogenesis of PN is not well understood, it is believed to involve cross talk between sensory nerve fibers, fibroblasts, epidermal keratinocytes and immune cells which results in the development of a sustained itch-scratch cycle (1). While the histopathological changes that include epidermal hyperplasia and hyperkeratosis, dermal fibrosis, hypervascularization and an inflammatory infiltrate in the dermis have been well characterized (2), limited information exists on the neuro-immunological axis in PN pathology.
Previous reports have highlighted increased dermal nerve fiber density and a decreased intraepidermal nerve fiber density, indicative of a central role for neural dysregulation (1), and histological studies have characterized the cellular players in PN lesions, with T cells, mast cells and granulocytes (primarily neutrophils and eosinophils) being the most commonly implicated immune cells (2–4). However, beyond this characterization, the exact role of these cell types in driving PN pathogenesis is unclear. For example, Johansson et al. (3) demonstrated an increased frequency of eosinophils in proximity to cutaneous nerves in PN lesional skin; while a cross talk between eosinophils and cutaneous nerves is suggested, the exact pathogenic pathways have yet to be elucidated. Several molecular factors have been implicated in driving PN pathogenesis. These include eosinophil cationic protein (ECP) and eosinophil derived neurotoxin (EDN) (1), as well as neuropeptides such as calcitonin gene-related peptide (CGRP) and substance P (SP). Among the cytokines, interleukin (IL)-31 and oncostatin M (OSM) have been shown to have increased expression in PN lesions (5, 6), with most of these studies implicating IL-31, a neuroimmune cytokine and its receptor IL-31RA, in PN pathogenesis (5, 7). Indeed, a recent study by Hashimoto et al. (6) shows that the increased expression of IL-31 and IL-31RA (and OSM) in PN lesions correlates with itch intensity. IL-31 has further been shown to have a pro-inflammatory effect on keratinocytes through activation of STAT3 signaling (8), a signaling pathway implicated in keratinocyte proliferation (9). The recent Phase II study results for nemolizumab, a humanized anti-IL-31RA antibody, provided clear evidence that IL-31 is a major player in driving PN pathogenesis, with nemolizumab treatment effectively reducing both pruritus and severity of skin lesions in PN patients (10, 11).
In order to better characterize the molecular and cellular components driving PN pathogenesis as well as to elucidate the mode of action of nemolizumab in PN, we performed transcriptomic analyses on skin biopsies from patients enrolled in the above-mentioned study (10).
METHODS
Patient cohort
The current study is part of the Phase 2 trial of nemolizumab in patients with moderate to severe PN (10). Briefly, 70 patients were randomized to placebo (36 patients) or nemolizumab (34 patients), the latter at a dose of 0.5mg/kg body weight administered at baseline, week 4 and week 8. Peak pruritus score on the numerical rating scale (PP-NRS) was recorded: severity of pruritus on the numerical rating scale ranged from 0 (no itch) to 10 (worst itch imaginable) and the peak pruritus was estimated using the worst scores every 24 hours in a 7-day period, with the highest score recorded as peak score. Primary outcome of the study was change in PP-NRS from baseline. Improvement of PN lesions was also assessed by Investigator Global Assessment (IGA) score ranging from 0 (Clear: No nodules and no activity signs) to 4 (Severe: Generalized nodules, dome-shaped with activity signs).
Reconstructed Human Epidermis (RHE) model
The three-dimensional RHE models were generated as described previously (12–14). Briefly, RHE cultures were generated using Normal Human Dermal Fibroblasts (NHDF) and Normal Human Epidermal Keratinocytes (NHEK). RHE cultures used were full thickness, with a dermis and an epidermis, and composed of autologous Fibroblasts and Keratinocytes. They were scaffold-free (no exogenous matrix) allowing self-assembly of the different layers of the skin by the cells and avoiding some potential inhibitors contained in collagen matrix for the RNA extractions. The RHE were cultivated in 12 well plates with insert (1.2 cm2) for 42 days to obtain a minimum of two layers of dermis and six layers of epidermis. RHE cultures from 6 different donors were left unstimulated or stimulated with different cytokines or cytokine combinations (3 replicates per donor per condition): IL-31, IL-13, IL-17A, IFNγ, IL-31+ IL-13, IL-31+ IL-17A, IL-31+ IFNγ and IL-13+ IL-17A. Concentrations for each of the cytokines used were: IL-31 (500ng/mL), IL-13 (100ng/mL), IL-17A (200ng/mL), IFNγ (50ng/mL). RHE cultures were lysed 72 hours after stimulation, and RNA extraction was performed using MagMAX mirVana Total RNA isolation from Tissue Kit (ThermoFisher Scientific). RNA was purified and concentrated using RNA Clean & Concentrator-5 kit (Zymo Research), according to the protocol, and RNA concentrations ranged from 3 to 380 ng/μL. Total RNA was quantified using the QuantiFluor One RNA kit (Promega) on the GloMax-Multi+ Detection System (Promega). Library preparation was performed using the Illumina® Stranded mRNA Prep Ligation kit (Illumina), according to manufacturer’s recommendations.
Skin biopsy processing and RNA isolation
Skin biopsies were collected from 16 placebo and 15 Nemolizumab treated subjects. Samples included lesional and non lesional biopsies at baseline as well as lesional samples after 12 weeks of treatment (placebo or nemolizumab). RNA was extracted from skin biopsies using Tripure Isolation Reagent (Sigma-Aldrich) according to manufacturer’s instructions. From these RNA samples, a DNase treatment was applied using the RNase-Free DNase kit (Cat No. 79254, Qiagen), followed by the RNeasy® MinElute® Cleanup Kit (Cat No. 74204, Qiagen). Total RNA was quantified using the QuantiFluor One RNA kit (Promega). RNA concentration obtained ranging from 4 to 20 ng/μl. Total RNA was qualified using Fragment Analyzer 5300 (Agilent) with the Agilent HS RNA Fragment kit (Agilent). The RNA Quality Number (RQN) obtained ranging from 1 to 6. Library preparation was performed using the SMARTer Stranded Total RNA-Seq Kit V2 - Pico Input Mammalian kit (TaKaRa). The primers for DEFB4B (Fisher Scientific, Hs00823638_m1), KRT16 (Fisher Scientific, Hs00373910_g1), IL1A (Fisher Scientific, Hs00174091_m1), KRT6C (Fisher Scientific, Hs00752476_s1) were used. RPLP0 was used as housekeeping gene (Fisher Scientific, Hs00420895_gH). The qPCR was run on 7900HT Fast Real-Time PCR System. For analysis of qPCR data, 0 values were replaced with the smallest value of the biomarker that was not zero. Log2 transformation was applied and a mixed model analysis with random subject effect was used. The REML algorithm was used for calculation of p-values.
RNA-Seq
Libraries were quantified using the QuantiFluor One dsDNA kit (Promega) and library analysis was performed using Fragment Analyzer 5300 (Agilent) with the Agilent HS NGS Fragment kit (Agilent). Following size selection with AMPure XP beads, shotgun libraries were sequenced using NextSeq (Illumina) on an Illumina NextSeq500 sequencer, in 2×75bp (High Output Kit v2, 150 cycles).
RNA-seq data processing
Following adapter trimming, sequence reads for 83 unique samples were aligned to the human genome (GRCh37) using STAR (version 2.5.2) (15). Gene (GENCODE v29) expression levels were then quantified with HTSeq (0.6.1) (16) using reads uniquely mapped to one genome location. Two RNA-seq samples were identified as outliers and 81 samples were used in subsequent analysis. Only genes with on average at least one read per sample were retained. DESeq2 (17) (under Bioconductor release 3.11) in R (4.0.0) was used for expression normalization, and we applied negative binomial distribution to model the expression level for differential expression analysis. For non-lesional vs lesional and the baseline vs week 12 comparison, individual effect was included as covariate; for the placebo vs nemolizumab comparison, age and gender were controlled. Multiple testing correction, i.e., False Discovery Rate (FDR) ≤ 5%, and |log2 Fold Change| ≥ 1 were used as criteria to declare significant differentially expressed genes (DEGs).
Cytokine, cell signature, and functional inference analysis
We assessed the degree of overlap between the most significant DEGs with the transcripts induced by cytokines in keratinocytes (defined by FDR≤10% and 1.5-Fold Change (FC)). The top 1,000 most significant DEGs from each of the differential expression comparison were used for fair comparison. For the comparison against epidermal compartments gene signature, we conducted scRNA-seq on the epidermal layer of skin biopsy and identified the top 50 marker genes for the basal, differentiated, and keratinized layers, respectively. We then investigated the effect size (in log2 FC) for each marker gene in each differential expression comparison. For transcription factor analysis, promoter region was defined as 5,000 base pairs upstream of the transcription start site, and we used the MEME suite (AME version 4.12.0) (18) to compute the enrichment statistic for the transcription factor binding. Hypergeometric test was used to construct gene ontology/pathway enrichment analysis.
Immunohistochemical staining (IHC)
For IHC, skin sections of clinically and histologically confirmed prurigo nodularis were treated with citrate buffer (pH6). Sections were then rinsed (Labline wash buffer DCS WL 583 C2500, 1:20 with distilled water) and transferred to a staining chamber. For each section, 100 μl of primary AK was added at specific dilution and incubated for 45 min at RT. After incubation, the sections were rinsed 2 × 3 min and incubated with 1 drop each of DCS Polylink HRP Kit PD000RP for 15 min at RT. After rinsing again 2 × 3 min, 1 drop of streptavidin HRP (DSC Polylink HRP Kit PD000RP) was added to each section and incubated for 15 min at RT. After incubation was complete, the sections were rinsed again and incubated for 15 min at RT in 3-amino-9-ethylcarbazole (AEC, AC 1310100 AEC 2 component kit DCS). Counterstaining of sections took place with hemalaun. The following antibodies and dilutions were used: 1:150 α-KRT16 (Zytomed), 1:4000 α-S100A7 (abcam).
Weighted gene correlation network analysis (WGCNA).
We used genes that were expressed in at least 20% of the samples in our dataset for the WGCNA (version 1.70–3). The “softPower” parameter was picked as the smallest value that achieve at least r2≥0.75. Spearman correlation was used to compute the correlation, and minimum module size was set as 100. Upon module merging, a height cut of 0.2 was used.
RESULTS
Prurigo nodularis is characterized by abnormal keratinocyte differentiation and immune activation.
After quality control, there were 31 PN patients with transcriptomic data in both lesional and non-lesional skin samples. Using false-discovery rate (FDR) of ≤ 10% and |log2| ≥ 1 as criteria, we identified 5,943 differentially expressed genes (DEGs) when comparing the uninvolved and lesional skin at baseline, of which 2,060 genes were increased and 3,874 were decreased (Figure 1A). Genes with the most robust increase included KRT6C (588-fold, FDR=8.2×10−80), DEFB4A (150-fold, FDR=1.1×10−12), and KRT16 (90-fold, FDR=1.9×10−52). Decreased genes included LCE5A (11-fold decrease, FDR=8.1×10−18) and AQP7 (7.9-fold decrease, FDR=2.6×10−17). The most prominent up-regulated cytokines included the IL36-family members: IL36A (6.8-fold, FDR=1.8×10−4) and IL36G (8.4-fold, FDR=3.9×10−25); IL20 family members: IL19 (5.1-fold, FDR=7.4×10−4), IL20 (3.5-fold, FDR=1.7×10−3), IL22 (2.7-fold, FDR=2.9×10−2), IL24 (5.8-fold, FDR=3.8×10−10), and IL26 (4.9-fold, FDR=3.3×10−3). Other factors included IL1A (4.7-fold, FDR=1.0×10−12), and IL1B (4.1-fold, FDR=3.7×10−6). The Th2 cytokines IL4 and IL13 did not reach significance but the IL4R was increased by 2.6-fold (FDR=6.3×10−19) (Supplementary Table 1). We then performed functional enrichment analyses on the DEGs to define the biological processes associated with PN skin. The most prominent gene ontology terms included: “cornified envelope” (FDR=1.5×10−12), “epidermal cell differentiation” (FDR=6.4×10−10), “keratinization” (FDR=1.6×10−12), “peptidase regulator activity” (FDR=1.1×10−4), “interleukin-4 and 13 signaling” (FDR=6.8×10−7), “interferon alpha/beta signaling” and “response to interferon gamma” (FDR=4.1×10−7, and FDR=4.1×10−6, respectively), “IL23 pathway” (FDR=2×10−5), and “mitotic metaphase and anaphase” (FDR=3.8×10−10) (Figure 1B; Supplementary Table 2). The enriched biologic processes are reflective of key aspects of PN, including keratinocyte hyperproliferation, altered epidermal differentiation, and inflammatory responses; by focusing on key expression modules in the lesional skin transcriptome, we revealed enrichment of the proliferative marker Ki67 (MKI67), the cell cycle gene CDKN1A, and inflammatory networks involving IL-1 and IL-36 (Figure 1C). Several key cytokines including DEFB4, KRT6C, KRT16 and IL1A were validated by QRT-PCR and IHC (KRT16, and S100A7) (Supplementary Figure 1).
Figure 1. PN is characterized by immune activation and abnormal keratinocyte differentiation.
Number of differentially expressed genes (DEGs) in PN lesional vs. non-lesional skin (n=62, FC≥2 or FC≤0.5, FDR≤0.1) (A). Enriched GO categories in PN lesional skin (−log (FDR) values are shown) (B). Literature-based gene network obtained from the top 1,000 DEGs in PN skin, generated using Genomatix Pathway System (GePS, www.Genomatix.de). The picture displays the top connected genes co-cited in PubMed abstracts in the same sentence linked to a function word (most relevant genes/interactions). The color represents fold change: orange represent the genes that are increased, and green represent the genes that are decreased in PN lesional vs. non-lesional skin. Critical nodes included the proliferation markers Ki67 (MKI67), IL-1 family members IL36G and IL1A, and CXCL8 and CDKN1A (C). Number of genes and within cluster correlation in the modules identified from the weighted gene co-expression gene networks (WCGNA) analysis of non-lesional and lesional PN skin (D). Top functions enriched in key co-expression modules (module #5, #6 and #8) from PN skin. The observed-expected enrichment is shown on the x-axis (E).
To gain a better understanding of the disease regulatory networks involved in PN skin we performed weighted gene co-expression network analysis (WGCNA)(19). We identified 20 co-expression modules in non-lesional skin and 10 clusters in lesional PN skin, and highlighted modules that have notable changes in gene-gene correlations between non-lesional and lesional skin (Figure 1D–E; Supplementary Table 3). Interestingly, we were able to assign distinct functions for these co-expressing gene modules, especially for the co-expression modules in PN lesional skin, with the most prominent involved immunological processes (module #8) including “immune-response” (FDR=1.8×10−47), “defense response” (FDR=1.2×10−39); cell proliferation (module #6) including “cell cycle” (FDR=2.9×10−94), “DNA metabolic process” (FDR=8.7×10−67); and epidermal processes (module #5), such as “epidermis development” (FDR 3.5×10−10), “keratinization” (FDR=1.7×10−6). (Figure 1E). Other notable findings were changes in “extracellular matrix” (FDR=1.16×10−59; module #2) and included genes such as MMP14, MMP16, COL1A1, COL1A2, and COL3A1, which were modestly elevated (FC ≥ 1.4; FDR ≤ 6×10−2) in the lesional skin, consistent with the association of PN with skin fibrosis (2).
Transcriptomic changes in PN lesion are enriched for keratinocyte and T-cell signatures
Clinical results of the nemolizumab Phase 2 study in patients with PN showed that nemolizumab resulted in a higher percentage of improvement of pruritus and skin lesions with overall good safety profile (10). At Week 12 the proportion of patients achieving 4-point reduction of weekly average of the PP-NRS was significantly higher in the nemolizumab group compared to placebo (52.9% versus 8.3%, p<0.001). At Week 12 the proportion of subjects achieving IGA Success (Defined as IGA 0 [Clear: No nodules and no activity signs] or 1 [Almost Clear: Rare single nodules, flattened]) was significantly higher in the nemolizumab group compared to placebo (20.6% versus 2.8%, p=0.02). At Week 18 the proportion of IGA 0–1 in the nemolizumab was even higher compared to placebo (38.2% vs 5.6%; p=0.001, binomial test).
We used an in silico approach (xCell (20)) to infer specific cell-type signatures for each non-lesional and lesional PN skin sample. Enrichment for transcriptomic signatures associated with epithelial cells and keratinocytes was observed (p<0.001 and p<0.0001, respectively) (Figure 2A). There was also increased prominence of Th2 associated signature (p<0.0001), consistent with the enriched GO categories for IL-4/IL-13 (Figure 1B). Other inflammatory signatures, such as macrophages (p<0.01), were more variable (Figure 2A). To address the relationship of PN with other hyperproliferative skin diseases that also have strong inflammatory signatures, we compared the PN transcriptome against that of atopic dermatitis (AD) and psoriasis. In a 3-way comparison, a large number of up- and down-regulated genes were shared between all three diseases (Figure 2B, C). The correlation of the effect sizes in the lesional skin was more pronounced between PN and psoriasis (Spearman’s correlation ρ=0.64) than between PN and AD (ρ=0.55). Genes that are commonly up-regulated in both psoriasis and PN include those participating in cytokine activity (CCL3, CXCL10, IFNG, IL12B, IL19, IL1B, IL20, IL22.) and keratinization (KRT16, KRT17, LCE3A, LCE3E, etc.). (Supplementary Table 4)
Figure 2. Enriched transcriptomic cellular signatures and overlap with psoriasis and AD.
Cell type inference analysis on non-lesional (NL) and lesional (L) PN skin samples using xCell. The color spectrum in the heatmap represents relative cellular signature across samples. The bar on the left side shows statistical difference in the enrichment between lesional and non-lesional PN skin with the colors representing different p-value thresholds (A). Comparison of PN associated DEGs against DEGs in Psoriasis (Pso) and atopic dermatitis (AD) for increased and decreased DEGs (B). Correlation analysis between the effect sizes in PN lesion versus those in psoriasis (Pso) and atopic dermatitis (AD). Spearman’s rank-order correlation was included. The genes significant in the x-axis, y-axis, and both axes are colored in red/blue/purple respectively (C).
Transcriptomic changes in PN skin with the IL-31 receptor inhibitor nemolizumab.
To assess the therapeutic effect of the IL-31 receptor (IL-31R) inhibitor nemolizumab, we performed RNA-seq on data from PN biopsies prior to and after 12 weeks of treatment, with placebo controls, in a double-blinded study. At baseline, there were 16 and 15 individuals in the placebo and nemolizumab groups, respectively; at week 12, we obtained lesional samples from 18 of the patients (11 placebo and 7 nemolizumab). Using principal component analysis (PCA) we observed the mixing of lesional skin samples from patients with PN at baseline between the two treatments groups (placebo vs nemolizumab). After 12 weeks of treatment, the nemolizumab group had a trend towards grouping with the baseline non-lesional skin group, but it was not observed for the placebo cohort (Figure 3A). Accordingly, this was accompanied by sample clustering using genes that are identified to be differentially expressed in the baseline non-lesional vs lesional skin comparison, with 6 out of the 7 nemolizumab samples (86%) in week 12 grouping with the baseline non-lesional skin samples, versus only 57% (4 out of 7) of the placebo group (Figure 3B). Notably, nemolizumab treatment led to normalization of greater number of PN associated DEGs compared to placebo for both genes upregulated in PN lesional skin (969 nemolizumab vs. 211 placebo) and genes downregulated in PN lesional skin (1,268 genes nemolizumab vs. 166 for placebo) (Figure 4A). This was also reflected in the correlation between placebo and nemolizumab treated DEGs with much greater overlap between nemolizumab treated group and PN, when compared to placebo vs. PN for both increased and decreased DEGs (Figure 4B). GO categories enriched amongst DEGs decreased only in the nemolizumab treated group by week 12 included “cell cycle” (FDR=5.6×10−14), “keratinocyte differentiation” (FDR=1.8×10−4), and “interleukin-4 and 13 signaling” (FDR=1.5×10−2), whereas none of these GO categories were found in the placebo-controlled group (Supplementary Table 2). Notable genes increased in PN skin and normalized by nemolizumab treatment included IL22, STAT1, STAT3, IL1B, IL1A, IL36G, DEFB4, and IL36A (Supplementary Table 1), with IL22, IL36A/G showing the most profound nemolizumab-specific effect. These data demonstrate that nemolizumab treatment normalized both epidermal hyperproliferation and differentiation, in addition to decreasing inflammatory responses.
Figure 3. Transcriptomic changes associated with the anti-IL31R inhibitor nemolizumab.
Principal component analyses (PCA) of the transcriptomic data from PN biopsies prior to and after 12-weeks of prospective placebo controlled, double blinded clinical trial with the anti-IL-31R inhibitor nemolizumab. Different colors represent different treatment groups, with lesional samples shown as triangles, and non-lesional skin shown as circles (A). Heatmap showing 2-way clustering (using genes differentially expressed between non-lesional vs lesional skin at baseline) for all samples (B).
Figure 4. Effect of nemolizumab on PN associated transcriptomic changes.
3-way Venn diagram of increased and decreased DEGs in PN skin and overlap with DEGs (compared to baseline) in the nemolizumab and placebo cohorts (A/B). Correlation analyses between different groups (PN baseline vs. placebo and nemolizumab DEGs) (Spearman’s rank-order correlation) (C).
Nemolizumab response is accompanied by decreased IL-31/Th2 responses in PN skin.
To address the effect of nemolizumab treatment on inflammatory responses in PN skin we interrogated nemolizumab and placebo treatment response against cytokine response signatures generated in RHE cultures as well as human epidermal rafts as previously described by our group. We observed a consistent decrease in IL-31 responses, either solitary, or in combination with other inflammatory cytokines, including the Th2 cytokine IL-13 or IL-17A (Figure 5A; Supplementary Figures 2,3), providing clearer evidence of blockade of the IL-31 pathway by nemolizumab. Notably, IL-17A response genes were enriched in PN skin, likely corresponding to specific downstream immunological cascade overlapping between psoriasis and PN (Supplementary Table 4). The IL17A mRNA expression was itself not significantly different in non-lesional vs lesional skin, nor in the nemolizumab treatment by week 12 (Supplementary Table 1), suggesting that while IL-17A is not a dominant cytokine in PN, IL-17A signatures can be downstream of the IL-31 signaling. In terms of cellular transcriptomic changes, we observed more robust decrease in the nemolizumab group including for Th1 and Th17 responses (Figure 5B).
Figure 5. Nemolizumab treatment leads to normalized epidermal differentiation and decreased IL-31/Th2 responses in PN skin.
Nemolizumab treatment led to decrease in IL-31 and IL-13 responses in PN skin compared to placebo, along with decreased expression of IL-17A response genes. Color spectrum represents -log10 p-value of the hypergeometric enrichment test (A). Immune cell signature was tested against genes that are i) up-regulated in baseline non-lesional vs lesional PN skin; ii) down-regulated in the placebo baseline vs week 12; iii) down-regulated in the nemolizumab group baseline vs week 12 comparisons. Color spectrum represents -log10 p-value of the hypergeometric enrichment test (B). Cross comparing transcriptomic responses in PN skin against cellular signatures obtained from single-cell data of healthy epidermis demonstrated that nemolizumab treatment led to normalization of epidermal gene expression related to the differentiated layer (KRT10+) of the epidermis, corresponding to normalization of epidermal differentiation. The different nomenclatures correspond to the different layers of the epidermis with “basal” corresponding to KRT5+ basal cells, KRT10+ “differentiated” corresponding to the spinous layer, and “keratinized” corresponding to the granular layers (FLG+) (C). Transcription Factor Binding Site (TFBS) of PN associated DEGs that greater normalization amongst nemolizumab down-regulated genes compared to placebo. The axes represent -log10 (p) values, and the color spectrum reflects enrichment p-value for TFBS results for up-regulated genes in the baseline non-lesional vs. lesional skin comparison (D).
To determine the tissue compartment that has the largest contribution to cellular response to anti-IL-31R blockade, we compared the transcriptomic data from placebo and nemolizumab groups against gene signatures for epidermal compartments obtained from single-cell RNA-seq data. Our results show that the basal keratinocyte (KRT14+) signature was increased (in terms of mean expression) in lesional PN skin and was restored to a similar degree in both placebo and the nemolizumab treated groups. In contrast, the spinous (differentiated) (KRT10+) signature in lesional PN skin (mean of the change ~1.5 fold compared to non-lesional skin) was only restored in the nemolizumab treated group but not in the placebo group by week 12 (Figure 5C). We then performed transcription factor binding site (TFBS) analyses to further understand the transcriptional regulators of the transcriptomic changes in PN and following placebo or nemolizumab treatment. The results demonstrated that binding sites that are enriched among genes up-regulated in the baseline lesional skin are more likely to be enriched among the nemolizumab down-regulated genes by week 12 (Figure 5D). The most significantly regulated transcription factors included EGR4 (p=4.5×10−6 and p=1.2×10−8 for enrichment in the promoters of up-regulated genes in PN lesional skin and for nemolizumab down-regulated genes, respectively), a member of the EGF family of zinc finger transcription factors; STAT3 (p=2.2×10−4 and p=2.5×10−5, respectively); and KLF16 (p=4.5×10−5 and p=2×10−5, respectively) (Supplemental Table 5).
Nemolizumab leads to significant decrease in IGA scoring.
The IGA score was associated with the transcriptomic data (Figure 6A), and while IGA was similar at baseline in both groups, there was consistent decrease in IGA score to reach IGA 0 or 1 in the nemolizumab group, indicating achievement of clear or almost clear, whereas there were very few IGA responders (by week 18) in the placebo group. The same trend was also observed when considering PP-NRS responders (Supplementary Figure 4). Notably, the measured distance between the PC1/PC2 components for the nemolizumab treated group was significantly less than that of the placebo group (p=0.036) (Figure 6B; Supplementary Figure 5), consistent with a therapeutic response.
Figure 6. Nemolizumab driven decrease in IGA scoring was accompanied by tighter clustering of PN samples on PCA analyses after 12 weeks of treatment.
The IGA value was superimposed on the PCA coordinates from the transcriptomic data. Nemolizumab group is shown as circles, whereas placebo group is shown as triangles. Baseline (top) and week 12 of treatment (bottom) are shown. The responder represents patients achieved almost clear skin by week 18 (i.e. IGA=0 or 1), and are showed in larger circles or triangles (A). Nemolizumab treatment led to tighter clustering of PN samples on PCA analyses compared to biopsy samples from the placebo treated cohort (B).
DISCUSSION
The data presented here are, to our knowledge, the first to provide a comprehensive view of the global transcriptomic changes in PN skin to reveal novel and important insights into the mechanism of action and efficacy of the anti-IL-31 receptor inhibitor nemolizumab. Notably, these data reflect on the transcriptomic level many of the hallmark changes observed in PN, including epidermal alteration, inflammatory response (Figures 2 and 5), fibrosis (2), and pruritus (Figure 6), and how these changes are reversed and normalized with nemolizumab treatment.
Given the marked epidermal hyperplasia in PN skin (2), many of the changes in gene expression in PN skin are related to abnormal keratinocyte proliferation and differentiation (Figure 1). These epidermal alterations accounted for most of the overlap of PN with both AD and psoriasis (Figure 2; Supplementary Table 4), both of which are also characterized by marked epidermal hyperplasia and altered epidermal differentiation (21). Notably, these changes demonstrated significant improvement in the nemolizumab treated group by week 12 of treatment, but not in the placebo group (Supplementary Table 2). Furthermore, consistent with the therapeutic effect of nemolizumab, we observed greatest effect of normalization on the differentiated (corresponding to the supraspinous and granular) layers of the epidermis (Figure 5C), likely reflective of decreased keratinocyte proliferation and restoration of normal epidermal differentiation.
Fibrosis is a feature of PN, characterized by deposition of vertically oriented collagen fibrils in the reticular and papillary compartments of the dermis (2). We found modules of genes involved in extracellular matrix biology to be enriched in PN skin (Supplementary Table 3), involving both collagen 1 and collagen 3 genes. COL1A1, COL1A2, and COL1A3 mRNA were increased in PN skin at baseline (1.7-, 1.44-, and 1.52-fold respectively, along with robust increases in matrix metalloproteinases (MMPs) such as MMP1 and MMP10 (124-fold and 55-fold increase respectively) (see Supplementary Table 1). Although, we observed minor changes in collagen expression with treatment, we observed normalization in the expression of the MMPs at week 12, suggesting a shift towards normal collagen homeostasis.
Our data also demonstrate how nemolizumab driven transcriptomic changes in PN correlate with improvement in PN lesions and pruritus. Chronic pruritus is a debilitating symptom of PN and has a profound impact on quality of life (22). The pathophysiology of PN still remains unclear but possible factors include the Th2 cytokines, IL-31, IL-4 and IL-13, major cytokines involved in atopic dermatitis (23), with decreased density of intraepidermal nerve fibers being shown to be reduced in both lesional and non-lesional skin (24). The data from the nemolizumab treatment is consistent with both scenarios contributing to itch. Thus, nemolizumab treatment leads to suppression of Th2 and IL-4/IL-13 responses in PN skin and decrease in expression of factors such as KLF16, which has been shown to inhibit neurite growth (25–27). In addition, nerve growth factor (NGF), which we confirm is increased in PN skin (28), is also normalized with nemolizumab treatment at week 12, and to a larger extent than in the placebo group (Supplementary Table 1). No changes were seen for the expression of CGRP or substance P (TAC1). These data are highly suggestive of a broad effect of nemolizumab on pruritus and may account for the long duration of pruritus improvement seen beyond the last dose of nemolizumab (>2 months) (10). However, it remains to be determined whether nemolizumab treatment restores neural dysregulation including changes related to structure and function of nerve fibers in PN lesions.
Importantly, and consistent with PN being a disease process driven by inflammation, immune responses including Th2 (IL-4/IL-13) response, and type I and type II IFN responses were prominent in PN skin (Figure 1). Th2 responses closely correlate with itch in diseases such as atopic dermatitis (23), a frequently predisposing condition to PN development in a subset of patients (29). Interestingly, anti-IL-31 receptor inhibition significantly decreased not only IL-31 responses in keratinocytes in PN skin (Figure 5A) but also led to decrease in Th2 and Th17 cellular signatures (Figure 5B), corresponding to decreased IL-13 and IL-17 responses in keratinocytes (Figure 5A). The enriched IL-17 responses in PN skin, as well as contribution from IL-36, likely account for the greater overlap of PN skin with plaque psoriasis, in contrast to AD (Figure 2B and 2C). A recent publication described increased IL-22 in PN lesions accompanied by increased frequency of gamma-delta T-cells, along with increased expression of Th17/IL-17 induced genes, although increased IL17A mRNA expression was not detected (30). Increased expression of IL22 in PN skin (2.7-fold increase, FDR-2.9×10−2) was confirmed along with slight increase in the expression of CD3G and CD3D indicative of gamma-delta T cells (30), and expression of these genes were normalized by nemolizumab treatment (see Supplementary Table 1). We also failed to observe significant changes in IL17A mRNA expression in our PN data. In addition, changes in both Th2 and Th17 responses with nemolizumab treatment, and to a lesser extent type II IFN responses, suggest that these cytokines act downstream of IL-31 in PN. Thus, these data demonstrate that nemolizumab treatment has a robust anti-inflammatory effect in PN. This anti-inflammatory response, which may be reflected in normalized expression of a broad range of pro-inflammatory genes, and signal mediators such as STAT1 and STAT3 in nemolizumab treated PN skin, is an additional mechanism, beyond the direct anti-itch effect of nemolizumab, by which it may normalize and improve manifestations of PN.
In summary, we provide in-depth characterization of the transcriptomic changes in PN skin and demonstrate the broad mechanisms of action of nemolizumab. These data demonstrate the broad therapeutic effect of anti-IL-31 receptor inhibition with nemolizumab on multiple aspects of PN pathogenesis, including epidermal differentiation, inflammatory responses, pruritus and extracellular remodeling, and confirms the critical upstream role of IL-31 in PN pathogenesis.
Supplementary Material
CLINICAL IMPLICATIONS.
Treatment of patients suffering from Prurigo Nodularis with the anti-IL-31RA agent nemolizumab effectively decreases IL-31 responses followed by normalization of markers of epidermal proliferation and inflammation including Th2/IL-13 and Th17/IL-17.
ACKNOWLEDGEMENT
This study was supported by Galderma. Dr. Gudjonsson and Lam C. Tsoi are supported by the National Psoriasis Foundation, Babcock Endowment Fund, the National Institute of Health (R01-AR069071 and R01-AR073196 (JEG), P30-AR075043 (JEG, LCT), K01-AR072129 (LCT), the A. Alfred Taubman Medical Research Institute (JEG, JMK), as well as the O’Brien Kidney Research Core Center (P30DK081943) (CCB).
COI Statement: JEG has served as a consultant to Almirall, BMS, Sanofi, AbbVie, Novartis, Eli Lilly, Pfizer, Galderma, and received research support from Almirall, Janssen, Novartis, Pfizer, BMS/Celgene, Timberpharma, and Galderma. SS has served as a consultant to Almirall, Bayer, Beiersdorf, Bellus, Bionorice, Cara Therapeutics, Celgene, Clexio, DS Biopharma, Galderma, Menlo Therapeutics, Novartis, Perrigo, and Trevi Therapeutics, Dermasence, Galderma, Kiniksa, Sanofi, Vanda Therapeutics. JMK has received Grant support from Q32 Bio, Celgene/BMS, Ventus Therapeutics, and Janssen. JMK has served on advisory boards for AstraZeneca, Eli Lilly, GlaxoSmithKline, Bristol Myers Squibb, Avion Pharmaceuticals, Provention Bio, Aurinia Pharmaceuticals, Ventus Therapeutics, and Boehringer Ingelheim. FHR, PF, FR, AL, CP, VJ, JKK are employees of Galderma. All other authors have nothing to disclose.
Abbreviations:
- AD
atopic dermatitis
- ECP
eosinophil cationic protein
- EDN
eosinophil derived neurotoxin
- CGRP
calcitonin gene-related peptide
- DEGs
differentially expressed genes
- FDR
false discovery rate
- IGA
investigator global assessment
- IFN
interferon
- IL
interleukin
- NHEK
normal human epidermal keratinocytes
- NRS
numerical rating scale
- OSM
oncostatin M
- PN
prurigo nodularis
- RHE
reconstructed human epidermis
- SP
substance P
- STAT
signal transducer and activator of transcription
- Th
T-helper
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES
- 1.Huang AH, Williams KA, Kwatra SG. Prurigo nodularis: Epidemiology and clinical features. Journal of the American Academy of Dermatology. 2020;83(6):1559–65. [DOI] [PubMed] [Google Scholar]
- 2.Weigelt N, Metze D, Stander S. Prurigo nodularis: systematic analysis of 58 histological criteria in 136 patients. J Cutan Pathol. 2010;37(5):578–86. [DOI] [PubMed] [Google Scholar]
- 3.Johansson O, Liang Y, Marcusson JA, Reimert CM. Eosinophil cationic protein- and eosinophil-derived neurotoxin/eosinophil protein X-immunoreactive eosinophils in prurigo nodularis. Arch Dermatol Res. 2000;292(8):371–8. [DOI] [PubMed] [Google Scholar]
- 4.Liang Y, Marcusson JA, Jacobi HH, Haak-Frendscho M, Johansson O. Histamine-containing mast cells and their relationship to NGFr-immunoreactive nerves in prurigo nodularis: a reappraisal. J Cutan Pathol. 1998;25(4):189–98. [DOI] [PubMed] [Google Scholar]
- 5.Sonkoly E, Muller A, Lauerma AI, Pivarcsi A, Soto H, Kemeny L, et al. IL-31: a new link between T cells and pruritus in atopic skin inflammation. J Allergy Clin Immunol. 2006;117(2):411–7. [DOI] [PubMed] [Google Scholar]
- 6.Hashimoto T, Nattkemper LA, Kim HS, Kursewicz CD, Fowler E, Shah SM, et al. Itch intensity in prurigo nodularis is closely related to dermal interleukin-31, oncostatin M, IL-31 receptor alpha and oncostatin M receptor beta. Exp Dermatol. 2021. [DOI] [PubMed] [Google Scholar]
- 7.Williams KA, Huang AH, Belzberg M, Kwatra SG. Prurigo nodularis: Pathogenesis and management. Journal of the American Academy of Dermatology. 2020;83(6):1567–75. [DOI] [PubMed] [Google Scholar]
- 8.Kasraie S, Niebuhr M, Baumert K, Werfel T. Functional effects of interleukin 31 in human primary keratinocytes. Allergy. 2011;66(7):845–52. [DOI] [PubMed] [Google Scholar]
- 9.Sano S, Chan KS, DiGiovanni J. Impact of Stat3 activation upon skin biology: a dichotomy of its role between homeostasis and diseases. J Dermatol Sci. 2008;50(1):1–14. [DOI] [PubMed] [Google Scholar]
- 10.Stander S, Yosipovitch G, Legat FJ, Lacour JP, Paul C, Narbutt J, et al. Trial of Nemolizumab in Moderate-to-Severe Prurigo Nodularis. N Engl J Med. 2020;382(8):706–16. [DOI] [PubMed] [Google Scholar]
- 11.Kwatra SG. Breaking the Itch-Scratch Cycle in Prurigo Nodularis. N Engl J Med. 2020;382(8):757–8. [DOI] [PubMed] [Google Scholar]
- 12.Chiricozzi A, Nograles KE, Johnson-Huang LM, Fuentes-Duculan J, Cardinale I, Bonifacio KM, et al. IL-17 induces an expanded range of downstream genes in reconstituted human epidermis model. PloS one. 2014;9(2):e90284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Hanel KH, Pfaff CM, Cornelissen C, Amann PM, Marquardt Y, Czaja K, et al. Control of the Physical and Antimicrobial Skin Barrier by an IL-31-IL-1 Signaling Network. J Immunol. 2016;196(8):3233–44. [DOI] [PubMed] [Google Scholar]
- 14.Rouaud-Tinguely P, Boudier D, Marchand L, Barruche V, Bordes S, Coppin H, et al. From the morphological to the transcriptomic characterization of a compromised three-dimensional in vitro model mimicking atopic dermatitis. Br J Dermatol. 2015;173(4):1006–14. [DOI] [PubMed] [Google Scholar]
- 15.Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29(1):15–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Anders S, Pyl PT, Huber W. HTSeq--a Python framework to work with high-throughput sequencing data. Bioinformatics. 2015;31(2):166–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, et al. MEME SUITE: tools for motif discovery and searching. Nucleic Acids Res. 2009;37(Web Server issue):W202–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. Bmc Bioinformatics. 2008;9:559. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol. 2017;18(1):220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Tsoi LC, Rodriguez E, Degenhardt F, Baurecht H, Wehkamp U, Volks N, et al. Atopic dermatitis is an IL-13 dominant disease with greater molecular heterogeneity compared to psoriasis. J Invest Dermatol. 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Todberg T, Zachariae C, Skov L. Treatment and Burden of Disease in a Cohort of Patients with Prurigo Nodularis: A Survey-based Study. Acta Derm Venereol. 2020;100(8):adv00119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Campion M, Smith L, Gatault S, Metais C, Buddenkotte J, Steinhoff M. Interleukin-4 and interleukin-13 evoke scratching behaviour in mice. Exp Dermatol. 2019;28(12):1501–4. [DOI] [PubMed] [Google Scholar]
- 24.Schuhknecht B, Marziniak M, Wissel A, Phan NQ, Pappai D, Dangelmaier J, et al. Reduced intraepidermal nerve fibre density in lesional and nonlesional prurigo nodularis skin as a potential sign of subclinical cutaneous neuropathy. Br J Dermatol. 2011;165(1):85–91. [DOI] [PubMed] [Google Scholar]
- 25.Moore DL, Apara A, Goldberg JL. Kruppel-like transcription factors in the nervous system: novel players in neurite outgrowth and axon regeneration. Mol Cell Neurosci. 2011;47(4):233–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Wang J, Galvao J, Beach KM, Luo W, Urrutia RA, Goldberg JL, et al. Novel roles and mechanism for Kruppel-like factor 16 (KLF16) regulation of neurite outgrowth and ephrin receptor A5 (EphA5) expression in retinal ganglion cells. J Biol Chem. 2016;291(41):21422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Galvao J, Iwao K, Apara A, Wang Y, Ashouri M, Shah TN, et al. The Kruppel-Like Factor Gene Target Dusp14 Regulates Axon Growth and Regeneration. Invest Ophthalmol Vis Sci. 2018;59(7):2736–47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Johansson O, Liang Y, Emtestam L. Increased nerve growth factor- and tyrosine kinase A-like immunoreactivities in prurigo nodularis skin -- an exploration of the cause of neurohyperplasia. Arch Dermatol Res. 2002;293(12):614–9. [DOI] [PubMed] [Google Scholar]
- 29.Iking A, Grundmann S, Chatzigeorgakidis E, Phan NQ, Klein D, Stander S. Prurigo as a symptom of atopic and non-atopic diseases: aetiological survey in a consecutive cohort of 108 patients. J Eur Acad Dermatol Venereol. 2013;27(5):550–7. [DOI] [PubMed] [Google Scholar]
- 30.Belzberg M, Alphonse MP, Brown I, Williams KA, Khanna R, Ho B, et al. Prurigo Nodularis is Characterized by Systemic and Cutaneous Th22 Immune Polarization. J Invest Dermatol. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.