Vaccinia virus-specific molecular signature in atopic dermatitis skin

Abstract

Background: Eczema vaccinatum (EV), a disseminated viral skin infection, is a life-threatening complication of vaccinia virus (VV) inoculation in patients with atopic dermatitis (AD) and is thought to be associated with a defective innate immune response. However, the precise mechanism or mechanisms and key factor or factors of EV are unknown.

Objective: Given that patients with psoriasis, another inflammatory skin disorder, are not susceptible to EV, we compared the global transcriptional response of skin to VV in healthy subjects, patients with psoriasis, and patients with AD, focusing on AD-specific genes. We hypothesized that differences in the transcriptional response to VV between patients with AD and patients with psoriasis or healthy subjects would identify a defective pathway or pathways that might be associated with the development of EV.

Methods: Gene expression profiling of sham-treated and VV-treated unaffected skin explants from patients with AD (n = 12), patients with psoriasis (n = 12), or healthy subjects (n = 13) were generated with U133_Plus2 (54,613 probe sets) GeneChips and analyzed with the GCOS_1.4/SAM_2.1/MAPPFinder_2.0 pipeline.

Results: Sixty-seven genes were significantly affected by VV in AD skin but not in psoriatic and healthy skin. Genes associated with defense response, response to wounding, and immune response were the most affected by VV in AD skin. All genes in these ontologies were downregulated, including the innate immunity genes leukotriene B(4) receptor (LTB4R), orosomucoid 1 (ORM1), coagulation factor II (thrombin) receptor (F2R), complement component 9 (C9), and LPS-binding protein (LBP). These findings were confirmed by means of real-time PCR and validated by means of PubMatrix analysis. ORM1, Toll-like receptor 4 (TLR4), and NLR family pyrin domain containing 1 (NLRP1) genes were also linked to AD severity.

Conclusion: This study identified groups of innate immunity genes that are associated with the aberrant response of AD skin to VV and represent potential targets for EV pathogenesis.

Figure 1. Atopic dermatitis-specific (A and B) and vaccinia virus -induced (C and D) transcriptional changes in skin biopsies from unaffected areas

Gene expression analysis was conducted using standard (solid circle) and clustering (dashed circles) approaches (see Materials and Methods). The “single” pool represents the number of significantly different genes that were detected using information from individual probe-set per gene (generate more than one value per gene). The “doublets” and “triplets” pools represent the number of significantly different genes that are targeted by two or three probe-sets, respectively and were analyzed using combined values of those probe-sets (generate only one value per gene).

Figure 2. Selection of first-line VV-specific gene candidates using cross-referencing approach

83 genes that were significantly affected by VV treatment in AD samples compared to both normal and psoriatic samples were cross-referenced with 106 genes that were significantly associated with AD in both comparisons to normal and psoriatic skin. The resulting pool of 67 genes represents VV-specific response in AD samples.

Figure 3. Expression level of VV-specific gene candidates in unaffected skin from AD patients detected by real time RT-PCR

The relative message abundance of 9 genes that are listed on the X axis was evaluated by real time RT-PCR and compared to internal controls (three housekeeping genes) as described in Materials and Methods using total RNA isolated from skin stimulated with VV in 3 randomly selected AD patients. The changes in transcript abundance were identified by comparing AD values to randomly selected VV-stimulated controls (n=3, solid bar) or psoriatic samples (n=3, gray bar) and expressed as mean fold changes with error bars representing standard deviations (SD). Significant changes (p<0.05) in transcript abundance are identified with asterisk (*). Full gene names are reported in Table III.

Figure 4. EASI-associated linear regression graphs of the most representative genes involved in defense response

Expression values (Y axis) of ORM1, TLR4, and NLRP1 genes were plotted against EASI score (X axis) and correlation coefficient (R) and two-tailed probability (P) calculated using Python Statistical package (see Methods). Fold change (FC) was calculated as the ratio between expression values corresponding to the three lowest and three highest EASI scores (dashed ovals). ORM1 - orosomucoid 1; TLR4 - toll-like receptor 4; NLRP1 - NLR family, pyrin domain-containing 1.