Dysregulation of inflammatory gene expression in lesional and nonlesional skin of patients with pyoderma gangrenosum

Dear Editor, Pyoderma gangrenosum (PG) is an inflammatory skin condition that often presents as chronic ulcers. It is often idiopathic but can be associated with systemic conditions.1 Our recent publications1,2 highlight the need for better characterization of the molecular understanding of PG to improve diagnostic accuracy and targeted therapy. Recent reviews suggest that innate immunity may have a causative and consequential role in PG. In support of this idea, we reported a possible role for pattern recognition receptors (PRRs) in the pathogenesis of PG.2 To test this hypothesis, paired skin biopsies were collected from lesional and nonlesional skin (> 5 cm from the affected area on the ipsilateral limb) in five patients with PG with clinically active disease. Freshly biopsied samples were immediately solubilized, lysed and purified for RNA. The resultant cDNA libraries were assessed for inflammatory gene expression of 84 genes using a commercially available Innate and Adaptive Immune Responses RT2 Profiler PCR Array (Qiagen, Venlo, the Netherlands). Amplification data were assessed for area under the curve (AUC) with a threshold of 0·90 to maximize the true positive rate, which revealed differential patterns of gene expression between lesional skin and nonlesional skin using the effect size delta. Of these, 49 genes demonstrated an AUC > 90 (Fig. 1). A multidimensional scaling plot confirmed clustering across the 49 genes with an AUC > 0·90. Differential gene expression in lesional vs. nonlesional skin in patients with pyoderma gangrenosum (PG). Gene array analysis detects differential gene expression between PG lesional skin (red) and nonlesional skin (blue). The effect size (δ) between the Ct values (threshold cycle for quantification of gene expression) of lesional and nonlesional skin was calculated: (mean in nonlesional skin) − (mean in lesional skin)/(SD in lesional skin). Data were expressed as fold differences relative to nonlesional skin. The effect size demonstrated a 1·2–2·8 difference relative to nonlesional skin in patients with PG. An additional healthy skin sample (green diamond) was collected from a person without PG as a comparator. The gene expression of the normal skin from the healthy control was comparable with that of the nonlesional skin in patients with PG. A reference gene was not chosen, because it has been documented that wounded and nonwounded skin have differences in housekeeping gene expression stability. A lower Ct indicates higher gene expression, and values ≥ 35 are considered as not detectable. Therefore, for undetected genes, a Ct value of 35 was used when a comparative calculation was indicated. AUC, area under the curve. + and × indicate mean values. Based on the Qiagen pathway annotation, a hypergeometric test found the strongest enrichment in the innate immunity pathways: (i) PRRs (P = 0·05) and (ii) other innate immunity genes (P = 0·05). Although other cytokines and chemokines were also overexpressed, they did not reach statistical significance as independent pathways. However, a few individual genes belonging to both the adaptive and humoral immune response pathways were upregulated and had statistical significance (P ≤ 0·05) with AUC > 0·90. These included STAT1, IL1R1, MAPK8, IRF3, NFKB1, MX1, HLA‐A, TLR4, TLR6, CD40, CD40LG, ITGAM and IRF7. To our knowledge this is the first study comparing lesional and nonlesional skin in patients with PG. The innate immune signalling pathway was found to be upregulated in lesional PG. Multifunctional genes have previously been reported as highly upregulated in lesional skin of PG: STAT1, STAT3, TYK2,3,CXCL8, CCL5, IL1B, IL1R1, CD40 and CD40LG.4 Previous studies have highlighted the overexpression of interleukin (IL)‐1β as a key cytokine in the pathogenesis of PG by driving autoinflammation and neutrophilic infiltration.4,5,6 However, elevated IL‐1β is not specific to PG as it is also overexpressed in clinical mimickers of PG, lessening its potential as a specific disease marker. Moreover, not all patients with PG respond to IL‐1 antagonists,5,7 necessitating further exploration for other possible pathways to provide more precise diagnosis and treatment options. The innate immune system is the sentinel for host defence and expresses proteins called PRRs, which include the membrane‐bound PRRs (Toll‐like receptors) and the cytoplasmic PRRs [NOD‐like receptors (NLRs) and RIG‐I‐like receptors (RLRs)]. Our data show PRRs to be upregulated in lesional vs. nonlesional skin in the same patient, suggesting a greater sensitivity in promoting an immune response. DDX58 is part of the RLR family that acts as a viral RNA sensor, essential for type I interferon induction. Mutations in these receptors have recently been linked to autoimmunity.8 Overexpression of Toll‐like receptors has been reported in diseases associated with PG such as inflammatory bowel disease and rheumatoid arthritis, suggesting a pathogenic role. Finally, NODs and NLRPs belong to the NLR group. Whereas NOD1 gene polymorphisms show conflicting results in patients with irritable bowel disease, NLRP3 has recently been reported as being associated with PG,6 and is consistently linked with Crohn disease, a disease classically associated with PG. On the other hand, there is increasing evidence that supports the presence of a dynamic cross‐talk between innate and adaptive immunity, with a pivotal role for pathways governing innate immune responses. Signal transducer and activator of transcription (STAT)3 (a T helper 17 marker) and STAT4 (a T helper 1 marker), which are involved in the generation and stimulation of IL‐6 and IL‐23, were elevated in our cohort. Monoclonal antibodies targeting these two interleukins, tocilizumab and ustekinumab, have been reported as therapeutically successful for some cases of PG. In addition, JAK2 (another adaptive immune gene) and STAT1 were found to be elevated. Interestingly, JAK2 stimulates STAT1, which has a dual function in the innate and adaptive immune system. The relevance of this may explain the therapeutic effectiveness of ruxolitinib, a Janus kianse (JAK)1 and JAK2 inhibitor, in PG. The main limitations of the study were the sample size, the homogenization of skin samples and the possibility that the results could be found in any type of inflammatory ulcer. In summary, this study confirms that genes implicated in immunity are upregulated in lesions of PG; specifically, PRRs are focally upregulated and other innate and adaptive immunity genes might be involved. We thank Jodi Lapidus, PhD, and Katrina Ramsey, MPH, from the Biostatistics and Design Program at OHSU and Theresa Lusardi from the Bioinformatics Core at OHSU for their special expertise on analytical methods for quantitative polymerase chain reaction and microarray results. Funding sources: Dean's Innovation Award granted of the Dean's Office at OHSU to A.G.O.L. Internal grant funding from VCU Dermatology. Conflicts of interest: none declared.