Molecular clustering of genes related to the atopic syndrome: Towards a more tailored approach and personalized medicine?

Atopy is the genetic predilection to produce specific immunoglobulin (Ig) E following exposure to allergens. This predisposition results in the development of atopic dermatitis (AD), food allergy (FA), asthma and allergic rhinitis (AR): the atopic syndrome [1]. The worldwide prevalence of these manifestations in children varies between 15–20%, 1–10%, 3–29% and 9–15%, respectively, and in adults from 1–3%, 3–4%, 2–12% and 7–42%, respectively [2‐6]. Atopic manifestations share a common mechanism involving allergen-specific IgE, which triggers the release of inflammatory mediators, like histamine, in the skin, gastrointestinal tract, lungs and nose. The course of these manifestations over time is characterized by the atopic march, generally starting with AD in infancy and followed by FA, asthma and AR later in childhood [7]. However, it is known that the atopic march not always follows the classic sequence and may occur at any age [8, 9]. Furthermore, not all atopic patients will develop the complete spectrum of atopic manifestations [7]. Despite its heterogeneous presentation, patients with atopic manifestations are mostly uniformly treated with topical or systemic immunosuppressive agents and/or antihistamines resulting in varying therapeutic responses as well [10‐13].

Subgroups of the atopic phenotype, termed endotypes, are possibly responsible for the differences in disease manifestations and treatment responses. These endotypes are the result of variations in physiologic, biologic, immunologic and/or genetic mechanisms [14]. Various genetic loci associated with both inflammation and multiple atopic manifestations have been identified in recent years based on genome-wide association studies (GWAS), showing common genetic mechanisms involved in the atopic syndrome [15‐24]. Nevertheless, the genetics of the atopic syndrome remain complicated for different reasons. For example, gene polymorphisms in different genes might cause the atopic syndrome independent of each other, and bearing a predisposing gene polymorphism does not necessarily result in development of the atopic syndrome [24]. The genetic complexity in the atopic syndrome possibly results in its heterogeneous clinical phenotype. Defining the endotypic profile of atopic patients in more detail contributes to determination of more homogeneous subclasses of patients. Subclasses are currently defined based on clinical and immunological characteristics, like the type of immune response involved [25]. However, stratification of atopic patients based on their genetic defect or polymorphism linked to their expression profile of immune cell lineages has not yet been investigated. This endotyping approach could be of interest as immune dysregulation may play an important role in the pathogenesis of the atopic syndrome. Interestingly, the atopic syndrome is a prevalent comorbidity in primary immunodeficiency diseases (PIDs), for example in hyper IgE syndrome (HIES), Comèl Netherton syndrome and immunodysregulation polyendocrinopathy enteropathy X-linked (IPEX) syndrome, which suggests that the atopic syndrome could be caused by a genetic defect in pathways that are also involved in these monogenic PIDs [26, 27]. This is supported by the hypothesis of autoallergy, in which atopy seems to stand at the boundary between allergy and auto-immunity, given the presence of IgE antibodies against self-proteins [28‐30].

Several gene-targeted and/or pathway-targeted treatment strategies for PIDs have recently been under clinical evaluation, which could be of clinical benefit in the atopic syndrome as well. Identification of genetic pathways for these targeted and personalized treatment modalities is therefore essential.

We hypothesize that subclasses within the atopic syndrome exist based on genes that act in the same molecular pathway. Additionally, genetic defects in pathways that cause PID might also be involved in the atopic syndrome.

Therefore, the aim of this study is to define subclasses within the atopic syndrome via molecular clustering of atopy-related genes based on their expression profiles of immune cell lineages. We first evaluated the overlap between atopy-related genes and monogenic PID genes. Secondly, we clustered the atopy-related genes based on their expression profile of immune cell lineages of healthy mice. Finally, we analyzed the pathways in which the atopy-related genes are involved.

This is the first study that describes clusters in the clinically heterogeneous phenotype of the atopic syndrome based on gene expression profiles of immune cell lineages of healthy mice. The overlap between atopy-related genes (n = 160) and monogenic PID genes (n = 278) was limited to 22 (5.3%) genes. We identified seven distinct clusters within the atopic syndrome based on the expression profiles of atopy-related genes. Functional pathway analysis of all known atopy-related genes resulted in identification of Th cell-mediated processes underlying the atopic syndrome.

The atopic syndrome is a prevalent comorbidity in a number of PIDs, suggesting that the atopic syndrome can be a symptom of PIDs and that immune dysregulation plays a role in the pathogenesis. Interestingly, the number of overlapping genes in this study was limited (5.3%) and did not belong to one PID category according to the IUIS phenotypic classification or immunologic component [32]. Nonetheless, the overlapping genes were bundled in just two of the seven atopy-related gene clusters (cluster D and F), which suggests that these endotypes of the atopic syndrome are associated with the predisposition to develop a PID. However, atopy-related mutations in these genes might differ from the disease-causing mutations of the PIDs.

Current literature reports nine PIDs to be possibly related to the atopic syndrome, including autosomal dominant HIES (AD-HIES; STAT3), autosomal recessive HIES (AR-HIES; DOCK8), Comèl Netherton syndrome (SPINK5), hypogammaglobulinemia, selective IgA deficiency (SIgAD), IgM deficiency, IPEX (FOXP3), chronic granulomatous disease (CGD; CYBA, CYBB, NCF1, NCF2 and NCF4), and phospholipase C gamma 2 (PLCG2) gene associated antibody deficiency and immune dysregulation (PLAID; PLCG2), and 28 additional genetic PID conditions [27, 37]. Only eight genes (STAT3, DOCK8, SPINK5, FLG, ARPC1B, PGM3, ERBIN and TYK2) were extracted from HGMD using the atopic phenotype search. Furthermore, only two of the 22 overlapping atopy-related and PID-related genes identified in this study were reported in literature to be involved in PIDs and the atopic syndrome [27]. The discrepancy between literature and HGMD could firstly be explained by the recent expansion of novel mutations derived from next generation sequencing (NGS). Secondly, the atopic manifestations in PIDs, as described in literature, might be an occasional finding and not related to the disease causing genes of PIDs. Thirdly, the heterogeneous course and presentation of the atopic syndrome may make it difficult to associate genetic mutations with atopic manifestations. Moreover, the infectious symptoms in PIDs might be a more prominent clinical feature than the atopic manifestations, which therefore could have resulted in a registration bias.

We found a low number of mutations in most atopy-related genes in human (six or less mutations in 157 of the 160 genes), suggesting that other phenomena contribute to the disease such as post-translational modifications. Alternatively, various genes that interact with environmental factors might be involved in the atopic syndrome, in which each gene contributes only to a small amount of the overall disease risk [38]. Furthermore, the differences between the clusters could indicate that immune regulation plays a role in the atopic syndrome next to underlying genetic mechanisms.

Strikingly, two of the identified clusters (D and F) have a completely opposite expression profile, both in lymphoid and myeloid cell lineages. An explanation for this phenomenon may be that both clusters share the same upstream regulator. Depending on a gain or loss of function mutation in this enhancer, the gene expression profile can be influenced by an agonist or antagonist of this regulator. By performing a functional pathway analysis of the atopy-related genes in only clusters D and F, we would explore the functional significance of these clusters. The analysis resulted in the pathways “T helper cell differentiation”, “TREM1 signaling” and “Th1 and Th2 activation pathway”, which is completely corresponding with the pathways involved in all atopy-related genes (data not shown). Therefore, we could unfortunately not differentiate between the functional significance of all atopy-related genes and those included in clusters D and F.

The identified Th cell-mediated pathway supports the hypothesis that changes in the immune system underlie and could be involved in the pathogenesis of atopy. In AD it has been previously described that acute skin lesions are characterized by Th2 cell infiltration with a shift towards predominantly Th1 cells in the chronic phase [39‐41]. In addition, asthma was reported as a Th2 cell-mediated diseases driven by allergen exposure [42]. Moreover, patients with FA and AR are characterized by allergen-specific Th2 cell-mediated responses showing that the obtained Th cell-pathways involved in the atopic syndrome are in agreement with these of the individual atopic manifestations [43‐45]. In most of our identified clusters (except clusters F, G, H, I and J) the atopy-related genes do not show increased expression in T lymphocytes (Fig. 3). Therefore, genes in these clusters might be expressed in immunologic cells that co-interact with T lymphocytes, including Th cells, or in cells that are progenitors of Th cells.

This study has some limitations. Firstly, we might have missed gene expression profiles of barrier cells as we could not include terms concerning the skin barrier in the phenotype search in HGMD. However, by using the terms “atopic dermatitis” and “eczema” we have identified important barrier genes, like COL6A5, FLG (subtypes), FLG2, and KLK7. Secondly, some discrepancies were found in the HGMD database. The genes from the atopic phenotype search did not completely overlap with the results from the search on atopy-related mutations per gene. Therefore, we identified atopic phenotypes per gene on the results of both searches. Thirdly, we clustered genes based on their expression profiles in the ImmGen dataset, which uses characterized immune cells of healthy mice. The gene expression profiles of immune cell lineages in healthy mice may not be identical to these in (atopic) human. This explains why we could not cluster all human atopy-related genes including FLG, which is an important atopy-related gene based on the number of atopy-related mutations (n = 62). Furthermore, the data from mice cannot directly be applied for subgrouping of the atopic syndrome in human. Therefore, large cohorts of patients with the atopy phenotype should be sequenced using NGS to investigate whether atopy clusters could be generated based on the gene expression profiles of immune cell lineages of atopic human. Identification of clusters of atopy-related genes by NGS potentially opens novel ways to select eligible patients for pharmaceutical studies and could predict therapeutic responses.

This study shows a model, using data of healthy mice, to define clusters of the atopic syndrome based on gene expression profiles of immune cell lineages. We identified seven distinct clusters within the atopic syndrome in which Th cell-mediated pathways were most often involved. This supports the hypothesis that both genetic mechanisms and immune dysregulation have a role in the pathogenesis the atopic syndrome. It also opens up the possibility for identification of novel therapeutic targets towards a more tailored approach and personalized medicine.