Epigenetic modifications are associated with mRNA and cytokine expression changes in chronic rhinosinusitis: a multiomics study from the United States

Objectives/hypothesis: Chronic rhinosinusitis (CRS) may be triggered by environmental insults. We hypothesized that CRS results from epigenetic modifications of host DNA from external insults, leading to downstream RNA/DNA gene expression changes and immuno-mechanical disruptions. We therefore performed a multi-omics study integrating epigenetic (DNA methylation), transcriptomic (mRNA), and proteomic (cytokine) data of CRS sinonasal tissue to visualize interactions amongst these modalities to study our hypothesis.

Methods: Sinonasal tissue was collected from 14 prospectively enrolled CRS and control subjects. Cytokine, mRNA transcriptome, and DNA methylome analysis were performed. Multi-omics analysis via joint dimensional reduction (JDR) was conducted.

Results: Multi-omics unsupervised clustering separated subjects into two distinct groups: one cluster of 9 CRS subjects and another with 3 controls and 2 non-eosinophilic CRSsNP subjects. DNA methylation, followed by mRNA expression, contributed most to cluster assignment. DNA methylation was the most significant data modality contributing to total variance on JDR. Cytokines critical in CRS (IL-5, IL-13, IL-10, IFNγ, IL-6) associated with hundreds of differentially methylated regions (DMRs) and mRNA. On conjoint analyses, common upstream DMRs and mRNAs were linked to cytokines IL-5 and IL-13, cytokines IL-10 and IFNγ, and cytokines IFNγ and IL-6, respectively.

Conclusions: Our results support the hypothesis that environmental insults may be significant drivers of CRS pathogenesis through epigenetic mechanisms that result in dysregulated mRNA transcription and cytokine expression. The most novel part of this study is our multi-omics approach that used integration of epigenetic (DNA methylation), transcriptomic (mRNA), and proteomic (cytokine) data to uncover insights into CRS pathogenesis; this is the first of its kind in CRS etiopathogenesis. The multi-omics analysis clearly separated clusters of control and CRS subjects, demonstrating its validity in future research. The study also identified interactions of methylated DNA, mRNA, and cytokines in CRS pathogenesis, highlighting novel molecules and pathways that may be potential therapeutic targets.

1 Introduction

A complex interaction of unfavorable environmental insults in the susceptible host has been postulated to disrupt normal homeostatic mechanisms in chronic rhinosinusitis (CRS) (1). Proposed external stressors (“the environment”) include microbial pathogens, microbiome dysbiosis, exposure to allergens, and air pollution (1). In addition, genetic susceptibility may be one of several host factors that result in disease. Even though familial clustering has long been reported, it is unclear whether familial clustering results from shared genes or shared environments, as identifiable monogenic alterations have not been identified in most CRS patients (2–5). In a large population-based study from Utah, U.S.A., 1,638 CRS with nasal polyposis (CRSwNP) and 24,200 CRS sans NP (CRSsNP) subjects were matched to random controls; 1st and 2nd-degree relatives were found to have a 4.1-fold and 3.3-fold elevated risk for CRSwNP, respectively. For CRSsNP, 1st and 2nd-degree relatives had a 2.4-fold and 1.4-fold risk, respectively (2). Interestingly, spouses of CRSsNP patients were also found to have a 2-fold increased risk of CRSsNP (2). In Sweden, Bohman, et al. (6) found that the prevalence of CRSwNP in relatives was 13.4% vs. 2.7% in controls; a relative risk of 4.9 in the first-degree relatives. These studies generate questions about the pathogenic roles of genes, shared environments, or both.

Epigenetics is the study of environmental influences on gene expression. Epigenetic studies are particularly helpful in disease states such as CRS, where multiple host or environmental factors may influence disease pathogenesis (7). External impact is modulated through mechanisms such as DNA methylation, histone modifications, non-coding RNAs, and alternative polyadenylation (APA) (8, 9). Epigenetic changes can notably persist and be passed to the progeny for 2–3 generations. Epigenetics may help explain both familial clustering and the increase in prevalence. Early studies on CRS epigenetics have shown several differences in DNA methylation between CRS and control tissue (10, 11). However, most of these are based in Asia, with only two of our previous studies being conducted in the United States. In this current study, our goal was to investigate the association, if any, of epigenetic modifications and mRNA transcriptomic and proteomic changes characterizing CRS. Transcriptomics analyzes RNA molecules, such as messenger RNA (mRNA), to understand gene expression (4, 12, 13) and has helped identify mechanistic pathways. However, most CRS transcriptomics studies have been conducted in Asia (14–17), with three studies incorporating non-Asian subjects (18–20) this is a relatively novel approach in North America, where population genetics and the environment differ (21, 22). When connecting transcriptomics with proteomics, studies have been divergent, reporting a correlation between CRS mRNA expression (transcriptome) and the proteome (17, 23, 24), as well as discordance (21) highlighting the need for further research using a multi-omics approach. Multiomics studies incorporate multiple modalities with large data sets through bioinformatics tools to help uncover complex relationships and interactions of biological processes at various levels, and can also help identify pathogenetic pathways that may not be apparent when studying each “omics” field individually. However, multiomics analyses involving the epigenome, transcriptome, and proteome (cytokine) have previously not been studied in CRS.

We hypothesize that external insults cause epigenetic modifications of host DNA, resulting in unfavorable DNA and associated RNA and protein expressions, which result in immuno-mechanical disruptions associated with CRS pathogenesis. We tested our hypothesis by performing multi-omics analyses integrating epigenetic (DNA methylation), transcriptomic (mRNA), and proteomic (cytokine) data of CRS sinonasal tissue. Our secondary goal was to visualize interactions amongst these modalities to uncover novel molecules and pathways with potential roles in CRS pathogenesis.

2 Methods

This study was conducted at a tertiary-level hospital in Arizona after approval from the institutional review board (IRB ID: 16-008609). Subjects were classified into controls and CRS based on nasal endoscopy and sinus CT according to 2015 consensus guidelines from the American Academy of Otolaryngology-Head and Neck Surgery (25). CRS subjects were further classified into CRSwNP and CRSsNP. Patients on systemic corticosteroids, biological therapy, and systemic or topical antibiotics in the last 4 weeks were excluded, so as not to affect the baseline cytokine profile of sinonasal tissue. Control subjects were undergoing transsphenoidal endoscopic resection of pituitary adenoma and were negative on CT and endoscopy for sinusitis and had no nasal history suggestive of allergic rhinitis. Prospective data was collected on demographics, clinical diagnoses, and disease severity [patient reported 22-item sinonasal outcome test (SNOT-22) scores (26) and Lund Mackay (27) Sinus CT scores].

STATA BE/18.0 was used to assess any differences in age and sex distribution between the cohorts. The Mann–Whitney U-test was used to compare the difference in age distribution, and Fisher's exact test was used to compare the gender distribution between the cohorts. A p-value of <0.05 was chosen as the criterion of statistical significance. Sinonasal tissue samples for all multi-omics analyses were obtained at a single time point under direct endoscopic guidance for 11 CRS and 3 control subjects. Since sinonasal mucosal samples were obtained during surgery, standard surgical aseptic precautions were used. Specimens were stored at −80°C until analysis. Samples were placed into sterile 7 ml polycarbonate tubes (Sarstedt 71.9923.610) and frozen within 15 min in a −90°C bath of Novec-engineered fluid (3M HFE-7000) cooled in a HistoChill freezing bath (SP Scientific HC80A0). Ethmoidal tissue was used for DNA methylation and cytokine studies. RNA sequencing was performed on ethmoidal tissue in CRS patients and inferior turbinate tissue in controls per IRB approval. A part of the ethmoidal tissue was sent in formalin at the time of surgery for structured histopathology analysis as described by Snidvongs et al. (28) Subjects with tissue eosinophils ≥10 eos/hpf were classified as eosinophilic CRS (eCRS) and those with <10 eos/hpf as non-eosinophilic CRS (neCRS).

2.1 DNA methylation

DNA extraction was done using the QIAamp DNA Mini kit by Qiagen (Reference no. 51306). Reduced Representation Bisulfite Sequencing (RRBS) Library prep and Sequencing were done on Illumina's HiSeq4000. RRBS data were analyzed using a streamlined analysis and annotation pipeline (SAAP) for RRBS, SAAP-RRBS (29). Cytosine followed by a guanine nucleotide (CpG) loci were called differentially methylated CpGs (DMCs) when p ≤ 0.05 and the mean methylation difference for the CpG loci between groups was at least 5% (delta ≥5%). A requirement of having at least four CpG loci within a candidate differentially methylated region (DMR) was set. Further details are included in the Supplementary Section.

2.2 RNA-Sequencing

RNA samples underwent library prep using Illumina TruSeq® RNA Exome Library Prep kit (San Diego, CA). Libraries were sequenced in 2 pools per lane on an Illumina HiSeq 4,000 (100 × 2 paired-end reads) and base-calling using Illumina's RTA v2.7.7. Paired-end RNA sequencing reads were processed through the RNA-Seq bioinformatics pipeline, MAP-RSeq v3.1.4 (30). Differentially expressed genes (DEG) were identified from raw gene counts using edgeR 2.6.2 (31). DEGs were reported with log2 fold change and False Discovery Rate (FDR <5%). Canonical pathway analysis using Ingenuity Pathway Analysis (IPA) software (Ingenuity® Systems) identified significant pathways (p-value <5%). Further details are included in the Supplementary Section.

2.3 Cytokine analysis

Frozen specimens were weighed, thawed, mixed with phosphate-buffered saline (PBS) and protease inhibitors (Millipore Sigma, Burlington, MA), and homogenized. Supernatants were collected after centrifugation. Cytokine and chemokine levels (48-plex) were measured using a Millipore multiplex kit (Billerica, MA) on a Bio-Rad MAGPIX multiplex reader (Hercules, CA). The concentrations of cytokines were normalized to the concentration of total protein in each sample. Total protein was analyzed by using a BCA Protein Assay Kit (Thermo Fisher Scientific). The values of cytokines were divided by the values of total protein. Samples below the minimum detectable concentration (MinDC) were assigned half the MinDC, and values above the standard curve limit were assigned the highest standard. Cytokines and chemokines detected in <10% of samples (17) were excluded. Eosinophil peroxidase (EPX) levels were assessed using an in-house sandwich enzyme-linked immunosorbent assay (ELISA), like that described by Ochkur et al. (32).

2.4 Multi-omics analysis

2.4.1 Preprocessing

Batch effects were corrected using ComBat (33). Modality-specific normalization was followed by transformation and filtering to facilitate an equal contribution to the JDR model. Normalized-FPKM RNA-seq counts were log₁₀ + 1e⁻⁴ transformed to achieve a Gaussian distribution. To balance feature counts across modalities, cytokines (with the fewest features) were filtered along with methylation and RNA features, which were significantly different between CRS and controls (FDR <5%).

2.4.2 Modeling

Joint dimensionality reduction (JDR) was performed using the multi-omics factor analysis (MOFA) methodology to integrate data modalities and extract variability dimensions, called factors (34). The contribution of each modality to the variance explained by each factor was quantified. To determine the number of viable factors, a randomized dataset was used, and factors where this dataset contributed the most variation were disregarded. Default settings were used, with modifications to remove scaling between data modalities and to pre-scale the value ranges to ensure a more accurate comparison of feature loading weights between modalities.

2.4.3 Analysis

Hierarchical all-against-all (HAllA) clustering was used to link quantitative and categorical clinical variables to factors, identifying features driving factor-sample distributions. Joint pathway analysis or kinase enrichment inference was used to analyze contributing modalities and features after identifying the associating factor to the clinical variable of interest.

3 Results

Table 1 details the clinical characteristics of the subjects. No statistically significant differences were found in the age and sex distributions between CRS cases and control subjects.

Table 1

Table 1. Clinical characteristics of subjects.

3.1 The multi-omics approach was successful in separating CRS subjects from controls

Multi-omics unsupervised clustering separated CRS from Controls; DNA methylation modality most contributed to cluster assignment, followed by RNA transcripts.

Multi-omics unsupervised clustering revealed two distinct groups with clear separation (Figure 1A). Figure 1B depicts the clinical diagnosis of cluster constituents. Cluster 1 was found to be entirely constituted by CRS subjects (3 CRSwNP, 6 CRSsNP), and Cluster 2 included all 3 controls and 2 non-eosinophilic CRSsNP subjects. Figure 1C depicts cluster constituents based on tissue eosinophil status. Where all Cluster 2 constituents had <10 eos/hpf, 6 CRS subjects in Cluster 1 had high tissue eosinophilia (2 with >100 eos/hpf, and 4 with tissue eosinophils between 10 and 100 eos/hpf), and 2 had non-eosinophilic tissue. Next, we examined each cluster to identify associated pathways (Figure 1D, E). The known functions of the top pathways associated with Cluster 1 and Cluster 2 are depicted in Tables 2A, B, respectively. The modality that most contributed to cluster assignment was DNA methylation, followed by RNA transcripts (Figure 1F). DMRs and DE RNAs associated with both clusters were identified, the top 50 of which are listed in Tables 3A, B, respectively. Supplementary Tables S1A and S1B present the known functions of these genes.

Figure 1

Figure 1. (A) Multiomics clustering of samples considering all three data modalities, (B) constituents of each cluster by diagnosis (control, CRSwNP, CRSsNP), (C) constituents of each cluster by tissue eosinophil count, (D) significant pathways associated with cluster 1 and (E) cluster 2, (F) number of features from each data modality contributing to cluster assignment.

Table 2A

Table 2A. Pathways associated with cluster 1 (9 CRS subjects).

Table 2B

Table 2B. Pathways Associated with Cluster 2 (All 3 control and 2 CRS <10 eos/hpf subjects)

Table 3A

Table 3A. The Top 50 differentially methylated regions in DNA between cluster 1 and 2.

Table 3B

Table 3B. The Top 50 Differentially Expressed mRNA between Clusters 1 and 2

3.2 DNA methylation was the most significant data modality contributing to total variance

Tissue samples from 14 subjects demonstrated several unique features across DNA methylation, transcriptomic, and cytokine data. Figure 2 depicts the filtering and transformation strategy for each data modality. DNA Methylation was the most significant data modality contributing to Total Variance.

Figure 2

Figure 2. Filtering and transformation strategy for each data modality through volcano plots and histogram representations. Data value distributions were transformed to be as close to Gaussian as possible.

JDR resulted in five dimensions of variation (“factors”), which captured the most significant patterns of information (Figure 3B). Methylation was the most significant data modality contributing to total variance (Figure 3C). Factor 1 is mostly driven by methylation. Factors 4 and 5 have similar contributions from methylation and RNA expression. Factor 3 is mostly correlated with methylation and less with RNA expression. Factor 2's important contribution comes from cytokines; however, it was also moderately correlated to DNA methylation, and weakly to RNA expression (Figure 3B). Table 4 details significant pathways associated with each factor as referenced by studies of inflammatory mechanisms.

Figure 3

Figure 3. Total variance contribution from each data modality (methylation, RNA, cytokines) is shown: (A) total number of data points for each modality, (B) joint dimension reduction resulted in five dimensions of variation (“Factors”). The darker the color, the more the contribution to the individual factor from the modality (methylation, RNA, cytokine), (C) Modality contributing to the total variance.

Table 4

Table 4. Significant pathways associated with each of the five factors shown in figure 2B.

3.3 Correlation of factor variation with clinical features

The five factors of variation were correlated with clinical features using HAllA (Figure 4). Significant correlations for Factor 1 were with allergic rhinitis, absolute blood eosinophil count, tissue eosinophil counts, and clinical diagnosis. For Factor 2, the most significant correlations were allergic rhinitis and immune deficiency. Factor 3's correlations were tissue eosinophil counts, smoking, allergic rhinitis, and gender. Factor 4's most significant correlations were tissue eosinophil counts, allergic rhinitis, and clinical diagnosis. Factor 5's most significant correlations were allergic rhinitis, clinical diagnosis, tissue eosinophil counts, and pre-operative SNOT-22. Tissue eosinophil counts mostly correlated with Factors 4 and 3, and absolute blood eosinophil counts only significantly correlated with Factor 1. Allergic rhinitis strongly correlated with Factors 1, 2, 4, and 5, and moderately correlated with Factor 3. Age, previous sinus surgery, asthma history, total IgE, CT score, steroid nasal spray use, and AERD presented with weak correlations.

Figure 4

Figure 4. Hierarchical-all-against-all clustering used to represent the correlation between factor distribution and clinical metrics of samples. Similarly, behaving factors and clinical metrics are binned into clusters. Correlation is shown from high (blue) to low (beige).

3.4 Examination of sample distribution across factors: tissue eosinophilia was able to better cluster subjects in two distinct groups compared to phenotypic status

We examined sample distribution across all five factors identified with JDR. Figure 5 (A, B, C) shows the sample distribution colored by diagnosis, tissue eosinophils/hpf, and SNOT-22 scores, respectively. Whereas SNOT-22 seemed to lead to a random distribution, both clinical diagnosis (CRSwNP, CRSsNP vs. control) and tissue eosinophils were able to better cluster subjects in two distinct groups. This was especially evident for factor 4, where all 3 controls were separated from CRSwNP and/or CRS ≥10 eos/hpf.

Figure 5

Figure 5. Sample distribution across all five factors colored by (A) diagnosis/polyp status, (B) tissue eosinophil numbers/hpf, and (C) SNOT-22 scores.

3.5 DNA methylation and mRNA heatmaps failed to cluster CRSwNP and CRSsNP separately

The association of each cytokine with DNA methylation and mRNA expression in each subject was examined. DNA methylation (Figure 6A) and mRNA heatmaps (Figure 6B) showed all 3 control samples clustered together. There was no clear clustering observed for phenotypical subtypes of CRS by methylation and mRNA expression status, likely exposing the limitations of classifying only by polyp status.

Figure 6

Figure 6. Heatmaps of the association between each subject (x-axis) with (A) DNA methylation and (B) RNA expression on the y-axis. Color key for Subjects: CRSwNP is orange, CRSsNP is purple, and Controls are green. Within the heatmap, red represents higher hypermethylation and mRNA expression.

3.6 Correlation between DNA, RNA, and cytokine expression: Two distinct clusters of cytokines were noted, with opposed positive, neutral, and negative correlations for cytokines

Next, we investigated the correlation between DNA, RNA, and cytokine expression and identified two distinct clusters with opposed positive, neutral, and negative correlations for the cytokine-methylation analysis (Figure 7A). The first cluster included MCSF, FLT3l, GROa, RANTES, VEGFa, FGF2, EGF, sCD40l, PDGFAA, IP10, MIG, IL-18, MCP1, and IL-12p40. The second cluster included IL-4, IL-13, IL-5, IL-1RA, IL-8, IL-10, INFγ, IL-6, G-CSF, Eotaxin, MIP-1b, MIP-1a, Fractalkine, MDC, EPX, MCP3, and TGFα. Two distinct clusters of cytokines were noted with opposed positive, neutral, and negative correlations for cytokines-RNA expression analysis as well (Figure 7B). The first cluster included MCP3, MIP-1a, IP10, IL-18, TGFα, GROa, RANTES, FGF2, VEGFa, sCD40l, and PDGFAA. The second cluster included EGF, IL-4, IL-1RA, MDC, EPX, IL-5, IL-13, MIG, Fractalkine, MCP1, IL-12p40, MCSF, FLT3l, Eotaxin, IL-8, IL-6, GCSF, MIP-1b, IFNγ, and IL-10.

Figure 7

Figure 7. Heatmap of (A) differential DNA methylation and (B) differential RNA expression to show correlations to cytokines. The y-axis represents the genes associated with the DMRs/ mRNAs, and the x-axis represents individual cytokines. Red: strong correlation; blue: weak correlation; white: no correlation.

3.7 Associations of individual cytokines with upstream DNA methylation and RNA expression were found

Isolated cytokine analysis was used next to study associations between cytokine, DNA, and RNA. The analysis revealed that IL-5 was associated with 720 differentially expressed (DE) RNAs and 172 differentially methylated regions (DMRs) on the DNA. IL-13 associated with 49 DE-RNAs and 180 DMRs, IL-10 to 54 DE-RNAs and 82 DMRs, IFNγ to 71 DE-RNAs and 123 DMRs, and IL-6 to 236 DE-RNAs and 178 DMRs. IL-4 and TGF did not significantly correlate with the other data modalities. Figures 8A,B illustrate the top 50 differentially methylated genes and differentially expressed mRNAs, respectively, for each of the 30 cytokines. Table 5 presents the top 10 DMRs and the differentially expressed mRNA identified in our study as related to many of these cytokines.

Figure 8

Figure 8. Heatmap correlating each cytokine (x-axis) with their top 50 genes with (A) differential DNA methylation & (B) differentially expressed RNA (y-axis).

Table 5

Table 5. Top 10 differentially methylated DNA as well as differentially expressed mRNA molecules.

3.8 Conjoint cytokine analyses identified common upstream DNA methylation and RNA expression for some cytokines

Next, conjoint cytokine analysis was performed (Figure 9) to identify commonly shared genes. The conjoint analysis showed that cytokines IL-5 and IL-13 were similarly correlated with RNA expression of TMEM74B and CPNE7, and with DNA methylation of DICER1 and SHISAL1. IL-10 and IFNγ were correlated to RNA expression of CYP27C1 and SOX18, and with DNA methylation of CASZ1, SYNRG, SNORD149, NTF4, and CTBP2. IFNγ and IL-6 were similarly correlated to RNA expression of CD79B and GFBP3, and with DNA methylation of EGFL7, HOXA2, PEBP4, WNT7B, CTBP2, and INTS1 (Figure 9). Table 6 enlists the function of genes identified on conjoint cytokine analysis.

Figure 9

Figure 9. Summary map relating cytokine to RNA expression, and RNA expression to DNA methylation.

Table 6

Table 6. Function of genes identified on conjoint cytokine analysis.

4 Discussion

The results of our study support our hypothesis that environmental insults may be significant in CRS pathogenesis through epigenetic mechanisms that result in dysregulated mRNA transcription and cytokine production downstream. Chronic dysregulated immune responses may continue long past the initial external insult through the induction of epigenetic changes, as seen in CRS (5).

Although the changes that occur at the histopathological and cytokine/protein levels have recently become better characterized in subjects with CRS (35–37), the genetic mechanism associated with such changes has not been fully characterized (3). In a sparse area of research, this study provides the first multi-omics analysis of CRS tissue from the United States, validating the association of epigenetic changes with transcriptomic and proteomic signatures seen in CRS. Furthermore, multi-omics analysis using DNA methylation, mRNA expression, and cytokine expression datasets successfully separated clusters of control and CRS subjects, demonstrating the utility of multi-omics analysis as a valuable tool in studying CRS. Our study is novel in using a multi-omics integration of DNA, RNA, and cytokine data to study CRS. Only two prior multi-omics studies have investigated CRS, but neither studied DNA data, and both were conducted outside of North America (18, 38). Miyata et al. (38), isolated eosinophils from six nasal polyp patients and performed multi-omics analysis using lipidomics, proteomics, and transcriptomics. Hoggard et al. (18), investigated temporal changes in polyp tissue in CRS in response to systemic corticosteroids in three males with CRSwNP subjects who underwent surgery, assessing natural variability over time and local response to systemic corticosteroid therapy. The authors found that the most highly abundant transcripts and proteins were associated with pathways involved in inflammation, FAS, cadherin, integrin, Wnt, apoptosis, cytoskeletal signaling, coagulation, and B- and T-cell activation. Given that DNA methylation was the most significant data modality contributing to the total variance between CRS and control subjects, epigenetic modifications are critical for further study in CRS for mechanistic and therapeutic targets. In addition, epigenetic mechanisms help explain shifts in the dominant CRS inflammatory pattern from non-type 2 to type 2, as is being noted in Asian regions as they undergo industrialization (36, 37, 39).

Our study further identified several known and potential mechanistic pathways and proteins involved in immunity and structural integrity, which may have roles in CRS pathogenesis. These are related to cytokine signal transduction, granule fusion events, phagosome maturation, toll-like receptors (TLRs) activation, reactive oxygen species formation, cellular metabolism, translational regulation, etc. The identification of JAK signaling also highlights the potential therapeutic role of JAK inhibitors in recalcitrant CRS, like current trials for asthma therapy (40). Many novel DMRs and DE mRNA (Tables 4A, B) were identified, including genes involved in membrane stability, homeostasis, as well as the gustation pathway, which are targets for further research (Supplementary Table S1).

Novel findings on conjoint cytokine analysis (Figure 9) showed that the cytokines IL-5 and IL-13 shared genes with RNA expression of TMEM74B and CPNE7, and with DNA methylation of DICER1 and SHISAL1. We also similarly noted shared genes for IL-10 and IFNγ, as well as IFNγ and IL-6. Table 6 details the functions of these genes. Both IL-5 and IL-13 are well recognized for their roles in the type-2 inflammatory process predominantly associated with CRSwNP, and hence, their association in differentially regulated upstream DNAs and RNAs is understandable, further reinforcing the utility of the multiomics approach. IFNγ is detected at lower levels in CRS tissue, reducing the antiviral immune response, which could result in or exacerbate the CRS following a viral infection (41). The role of IL-10 and IL-6 is reported in the literature (42, 43), and their association with IFNγ is interesting. DMRs and DE RNAs identified in association with key inflammatory cytokines involved in CRS pathogenesis, like IL-5, IL-13, IL-10, IFNγ, and IL-6 (Table 5), could be potentially important future therapeutic targets.

4.1 Limitations

This is a small study, albeit with the largest number of subjects published for CRS multiomics. While the multiomics approach distinguished two clusters, one of which was composed entirely of CRS patients, the other grouped three controls and two non-eosinophilic CRSsNP subjects, and perhaps these may have been clustered differently in a larger sample size. Genomic assays and multiomics analysis are prohibitively expensive, complex, and require technical expertise in integration, statistics, and systems biology. However, we hope that with reduced cost of genomic assays and multiomics analysis and support from extramural funding, larger prospective sampling can be performed for future studies.

Prospective, longitudinal studies with sample collection at multiple time points are needed to study CRS disease evolution. RNA expression is transient and may not correlate with protein level unless analyzed concurrently, which was mitigated by collecting tissue for histopathology, DNA methylation, and cytokine assay simultaneously in this study.

The multi-omics approach may also allow for focused upstream gene profiling of targeted cytokines of interest, such as IL-4, IL-5, IL-13, and others, in addition to an unsupervised approach that was used in this study. We anticipate that the use of single-cell RNA sequencing may be necessary for this approach rather than the bulk tissue sample that was used for this study.

Technical limitations in the study include bulk tissue RNA sequencing, which can only provide an average gene expression profile for the entire sample, but is cheaper than single-cell RNA sequencing (scRNA-seq) while identifying global differences in gene expression between disease and control states. Additionally, of the multiplex assay performed for cytokines, only 30 could be included per study methodology for multiomics analysis. Quantifying tissue eosinophilia with histopathology is imperfect, as degranulated eosinophils are difficult to measure. More sensitive novel assays, such as those from NanoString Technology (https://nanostring.com/), are planned for future study.

5 Conclusions

The study supports the hypothesis that environmental insults may be significant drivers of CRS pathogenesis through epigenetic mechanisms that result in dysregulated mRNA transcription and cytokine expression. The most novel part of this study is the integration of epigenetic (DNA methylation), transcriptomic (mRNA), and proteomic (cytokine) data to uncover novel insights into the pathogenesis of CRS. This multi-omics approach is the first of its kind to study environment-host interactions in CRS etiopathogenesis. The multi-omics analysis clearly separated clusters of control and CRS subjects, demonstrating its validity in future research. DNA methylation also contributed most to total variance, underscoring the role of environmental factors in CRS. Key cytokines like IL-5, IL-13, IL-10, IFNγ, and IL-6 were associated with hundreds of differentially methylated regions (DMRs) and differentially expressed mRNAs, providing future targets for study. IL-5 and IL-13, IL-10 and IFNγ, and IFNγ and IL-6 were associated with common upstream genes. The study further identified interactions of methylated DNA, mRNA, and cytokines in CRS pathogenesis, highlighting novel molecules and pathways that may be potential therapeutic targets.

Data availability statement

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

Ethics statement

The studies involving humans were approved by Institutional review Board, Mayo Clinic, Arizona, US. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

DL: Project administration, Writing – original draft, Methodology, Resources, Visualization, Conceptualization, Validation, Writing – review & editing, Funding acquisition, Supervision. TB: Writing – original draft, Writing – review & editing. CM: Formal analysis, Data curation, Software, Investigation, Writing – review & editing. EJ: Investigation, Software, Writing – review & editing, Data curation, Formal analysis. NK: Writing – review & editing. PL: Writing – review & editing, Writing – original draft. MM: Writing – review & editing. AM: Writing – review & editing. HK: Conceptualization, Supervision, Writing – review & editing, Methodology.

Funding

The author(s) declare that financial support was received for the research and/or publication of this article. This work was partly funded by an unrestricted grant through the FlinnFoundation, and intramural grants from Mayo Clinic in Arizona.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declare that no Generative AI was used in the creation of this manuscript.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/falgy.2025.1606255/full#supplementary-material

References

1. Kato A, Schleimer RP, Bleier BS. Mechanisms and pathogenesis of chronic rhinosinusitis. J Allergy Clin Immunol. (2022) 149(5):1491–503. doi: 10.1016/j.jaci.2022.02.016

PubMed Abstract | Crossref Full Text | Google Scholar

2. Oakley G, Curtin K, Orb Q, Schaefer C, Orlandi R, Alt J. Familial risk of chronic rhinosinusitis with and without nasal polyposis: genetics or environment. Int Forum Allergy Rhinol. (2015) 5(4):276–82. doi: 10.1002/alr.21469

PubMed Abstract | Crossref Full Text | Google Scholar

3. Brar T, Marks L, Lal D. Insights into the epigenetics of chronic rhinosinusitis with and without nasal polyps: a systematic review. Front Allergy. (2023) 4:1165271. doi: 10.3389/falgy.2023.1165271

PubMed Abstract | Crossref Full Text | Google Scholar

4. Lal D, Brar T, Ramkumar SP, Li J, Kato A, Zhang L. Genetics and epigenetics of chronic rhinosinusitis. J Allergy Clin Immunol. (2023) 151(4):848–68. doi: 10.1016/j.jaci.2023.01.004

PubMed Abstract | Crossref Full Text | Google Scholar

5. Brar T, Marino MJ, Lal D. Unified airway disease: genetics and epigenetics. Otolaryngol Clin North Am. (2023) 56(1):23–38. doi: 10.1016/j.otc.2022.09.002

PubMed Abstract | Crossref Full Text | Google Scholar

6. Bohman A, Juodakis J, Oscarsson M, Bacelis J, Bende M, Naluai ÅT. A family-based genome-wide association study of chronic rhinosinusitis with nasal polyps implicates several genes in the disease pathogenesis. PLoS One. (2017) 12(12):e0185244. doi: 10.1371/JOURNAL.PONE.0185244

PubMed Abstract | Crossref Full Text | Google Scholar

7. Rakyan VK, Down TA, Balding DJ, Beck S. Epigenome-wide association studies for common human diseases. Nat Rev Genet. (2011) 12(8):529–41. doi: 10.1038/nrg3000

PubMed Abstract | Crossref Full Text | Google Scholar

8. Martin MJ, Garcia-Sanchez A, Estravis M, Gil-Melcón M, Isidoro-Garcia M, Sanz C, et al. Genetics and epigenetics of nasal polyposis: a systematic review. J Investig Allergol Clin Immunol. (2021) 31(3):196–211. doi: 10.18176/jiaci.0673

PubMed Abstract | Crossref Full Text | Google Scholar

9. Allis CD, Jenuwein T. The molecular hallmarks of epigenetic control. Nat Rev Genet. (2016) 17(8):487–500. doi: 10.1038/NRG.2016.59

PubMed Abstract | Crossref Full Text | Google Scholar

10. Brar T, Baheti S, Marino MJ, Kita H, Lal D. Genome-wide epigenetic study of chronic rhinosinusitis tissues reveals dysregulated inflammatory, immunologic and remodeling pathways. Am J Rhinol Allergy. (2023) 37:692–704. doi: 10.1177/19458924231193526

PubMed Abstract | Crossref Full Text | Google Scholar

11. Brar T, Baheti S, Bhagwate AV, Kita H, Marino MJ, Lal D. Genomewide epigenetic study shows significant DNA hypermethylation in chronic rhinosinusitis versus control ethmoidal tissue. Int Forum Allergy Rhinol. (2023) 13:2235–9. doi: 10.1002/alr.23205

PubMed Abstract | Crossref Full Text | Google Scholar

12. Wang Z, Gerstein M, Snyder M. RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet. (2009) 10(1):57–63. doi: 10.1038/NRG2484

PubMed Abstract | Crossref Full Text | Google Scholar

13. Poole A, Urbanek C, Eng C, Schageman J, Jacobson S, O'Connor BP, et al. Dissecting childhood asthma with nasal transcriptomics distinguishes subphenotypes of disease. J Allergy Clin Immunol. (2014) 133(3):670–8. doi: 10.1016/J.JACI.2013.11.025

PubMed Abstract | Crossref Full Text | Google Scholar

14. Wang W, Gao Z, Wang H, Li T, He W, Lv W, et al. Transcriptome analysis reveals distinct gene expression profiles in eosinophilic and noneosinophilic chronic rhinosinusitis with nasal polyps. Sci Rep. (2016) 6(May):1–14. doi: 10.1038/srep26604

PubMed Abstract | Crossref Full Text | Google Scholar

15. Gill AS, Pulsipher A, Sumsion JS, Oakley GM, Leclair LW, Howe H, et al. Transcriptional changes in chronic rhinosinusitis with asthma favor a type 2 molecular endotype independent of polyp Status. J Asthma Allergy. (2021) 14:405–13. doi: 10.2147/JAA.S301825

PubMed Abstract | Crossref Full Text | Google Scholar

16. Plager DA, Kahl JC, Asmann YW, Nilson AE, Pallanch JF, Friedman O, et al. Gene transcription changes in asthmatic chronic rhinosinusitis with nasal polyps and comparison to those in atopic dermatitis. PLoS One. (2010) 5(7):e11450. doi: 10.1371/JOURNAL.PONE.0011450

PubMed Abstract | Crossref Full Text | Google Scholar

17. Hao Y, Zhao Y, Wang P, Du K, Li Y, Yang Z, et al. Transcriptomic signatures and functional network analysis of chronic rhinosinusitis with nasal polyps. Front Genet. (2021) 12:609754. doi: 10.3389/fgene.2021.609754

PubMed Abstract | Crossref Full Text | Google Scholar

18. Hoggard M, Jacob B, Wheeler D, Zoing M, Chang K, Biswas K, et al. Multiomic analysis identifies natural intrapatient temporal variability and changes in response to systemic corticosteroid therapy in chronic rhinosinusitis. Immun Inflamm Dis. (2021) 9(1):90–107. doi: 10.1002/iid3.349

PubMed Abstract | Crossref Full Text | Google Scholar

19. Kenney HM, Rangel-Moreno J, Peng Y, Chen KL, Bruno J, Embong A, et al. Multi-omics analysis identifies IgG2b class-switching with ALCAM-CD6 co-stimulation in joint-draining lymph nodes during advanced inflammatory-erosive arthritis. Front Immunol. (2023) 14:1237498. doi: 10.3389/fimmu.2023.1237498

PubMed Abstract | Crossref Full Text | Google Scholar

20. Xu Z, Huang Y, Meese T, Van Nevel S, Holtappels G, Vanhee S, et al. The multi-omics single-cell landscape of sinus mucosa in uncontrolled severe chronic rhinosinusitis with nasal polyps. Clin Immunol. (2023) 256:109791. doi: 10.1016/j.clim.2023.109791

PubMed Abstract | Crossref Full Text | Google Scholar

21. Workman AD, Nocera AL, Mueller SK, Otu HH, Libermann TA, Bleier BS. Translating transcription: proteomics in chronic rhinosinusitis with nasal polyps reveals significant discordance with messenger RNA expression. Int Forum Allergy Rhinol. (2019) 9(7):776–86. doi: 10.1002/alr.22315

PubMed Abstract | Crossref Full Text | Google Scholar

22. Brar T, McCabe C, Miglani A, Marino M, Lal D. Tissue eosinophilia is superior to an analysis by polyp Status for the chronic rhinosinusitis transcriptome: an RNA study. Laryngoscope. (2023) 133:2480–9. doi: 10.1002/lary.30544

PubMed Abstract | Crossref Full Text | Google Scholar

23. Liu Z, Kim J, Sypek JP, Wang IM, Horton H, Oppenheim FG, et al. Gene expression profiles in human nasal polyp tissues studied by means of DNA microarray. J Allergy Clin Immunol. (2004) 114(4):783–90. doi: 10.1016/J.JACI.2004.04.052

PubMed Abstract | Crossref Full Text | Google Scholar

24. Stankovic KM, Goldsztein H, Reh DD, Platt MP, Metson R. Gene expression profiling of nasal polyps associated with chronic sinusitis and aspirin-sensitive asthma. Laryngoscope. (2008) 118(5):881–9. doi: 10.1097/MLG.0B013E31816B4B6F

PubMed Abstract | Crossref Full Text | Google Scholar

25. Rosenfeld RM, Piccirillo JF, Chandrasekhar SS, Brook I, Ashok Kumar K, Kramper M, et al. Clinical practice guideline (update): adult sinusitis. Otolaryngol Head Neck Surg. (2015) 152(2 suppl):S1–S39. doi: 10.1177/0194599815572097

PubMed Abstract | Crossref Full Text | Google Scholar

26. Hopkins C, Gillett S, Slack R, Lund VJ, Browne JP. Psychometric validity of the 22-item sinonasal outcome test. Clin Otolaryngol. (2009) 34(5):447–54. doi: 10.1111/J.1749-4486.2009.01995.X

PubMed Abstract | Crossref Full Text | Google Scholar

27. Lund VJ, Kennedy DW. Quantification for staging sinusitis. The staging and therapy group. Ann Otol Rhinol Laryngol Suppl. (1995) 167:17–21. doi: 10.1177/000348949510410s02

PubMed Abstract | Crossref Full Text | Google Scholar

28. Snidvongs K, Lam M, Sacks R, Earls P, Kalish L, Phillips PS, et al. Structured histopathology profiling of chronic rhinosinusitis in routine practice. Int Forum Allergy Rhinol. (2012) 2(5):376–85. doi: 10.1002/alr.21032

PubMed Abstract | Crossref Full Text | Google Scholar

29. Sun Z, Baheti S, Middha S, Kanwar R, Zhang Y, Li X, et al. SAAP-RRBS: streamlined analysis and annotation pipeline for reduced representation bisulfite sequencing. Bioinformatics. (2012) 28(16):2180–1. doi: 10.1093/BIOINFORMATICS/BTS337

PubMed Abstract | Crossref Full Text | Google Scholar

30. Kalari KR, Nair AA, Bhavsar JD, O'Brien DR, Davila JI, Bockol MA, et al. MAP-RSeq: mayo analysis pipeline for RNA sequencing. BMC Bioinformatics. (2014) 15:224. doi: 10.1186/1471-2105-15-224

PubMed Abstract | Crossref Full Text | Google Scholar

31. Robinson MD, McCarthy DJ, Smyth GK. Edger: a bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. (2010) 26(1):139–40. doi: 10.1093/BIOINFORMATICS/BTP616

PubMed Abstract | Crossref Full Text | Google Scholar

32. Ochkur SI, Kim JD, Protheroe CA, Colbert D, Condjella RM, Bersoux S, et al. A sensitive high throughput ELISA for human eosinophil peroxidase: a specific assay to quantify eosinophil degranulation from patient-derived sources. J Immunol Methods. (2012) 384(1-2):10–20. doi: 10.1016/j.jim.2012.06.011

PubMed Abstract | Crossref Full Text | Google Scholar

33. Johnson WE, Li C, Rabinovic A. Adjusting batch effects in microarray expression data using empirical Bayes methods. Biostatistics. (2007) 8(1):118–27. doi: 10.1093/biostatistics/kxj037

PubMed Abstract | Crossref Full Text | Google Scholar

34. Argelaguet R, Arnol D, Bredikhin D, Deloro Y, Velten B, Marioni JC, et al. MOFA+: a statistical framework for comprehensive integration of multi-modal single-cell data. Genome Biol. (2020) 21(1):111. doi: 10.1186/s13059-020-02015-1

PubMed Abstract | Crossref Full Text | Google Scholar

35. Bachert C, Zhang N, Hellings PW, Bousquet J. Clinical reviews in allergy and immunology endotype-driven care pathways in patients with chronic rhinosinusitis. J Allergy Clin Immunol. (2019) 141(5):1543–51. doi: 10.1016/j.jaci.2018.03.004

Crossref Full Text | Google Scholar

36. Karin J, Tim D, Gabriele H, Cardell LO, Marit W, Claus B. Type 2 inflammatory shift in chronic rhinosinusitis during 2007–2018 in Belgium. Laryngoscope. (2021) 131(5):E1408–14. doi: 10.1002/lary.29128

PubMed Abstract | Crossref Full Text | Google Scholar

37. Wang X, Zhang N, Bo M, Holtappels G, Zheng M, Lou H, et al. Diversity of TH cytokine profiles in patients with chronic rhinosinusitis: a multicenter study in Europe, Asia, and oceania. J Allergy Clin Immunol. (2016) 138(5):1344–53. doi: 10.1016/j.jaci.2016.05.041

PubMed Abstract | Crossref Full Text | Google Scholar

38. Miyata J, Fukunaga K, Kawashima Y, Watanabe T, Saitoh A, Hirosaki T, et al. Dysregulated fatty acid metabolism in nasal polyp-derived eosinophils from patients with chronic rhinosinusitis. Allergy. (2019) 74(6):1113–24. doi: 10.1111/ALL.13726

PubMed Abstract | Crossref Full Text | Google Scholar

39. Katotomichelakis M, Tantilipikorn P, Holtappels G, De Ruyck N, Feng L, Van Zele T, et al. Inflammatory patterns in upper airway disease in the same geographical area may change over time. Am J Rhinol Allergy. (2013) 27(5):354–60. doi: 10.2500/ajra.2013.27.3922

PubMed Abstract | Crossref Full Text | Google Scholar

40. Georas SN, Donohue P, Connolly M, Wechsler ME. JAK Inhibitors for asthma. J Allergy Clin Immunol. (2021) 148(4):953–63. doi: 10.1016/j.jaci.2021.08.013

PubMed Abstract | Crossref Full Text | Google Scholar

41. Hwang JW, Lee KJ, Choi IH, Han HM, Kim TH, Lee SH. Decreased expression of type I (IFN-β) and type III (IFN-λ) interferons and interferon-stimulated genes in patients with chronic rhinosinusitis with and without nasal polyps. J Allergy Clin Immunol. (2019) 144(6):1551–65.e2. doi: 10.1016/j.jaci.2019.08.010

PubMed Abstract | Crossref Full Text | Google Scholar

42. Kubota K, Takeno S, Taruya T, Sasaki A, Ishino T, Hirakawa K. IL-5 and IL-6 are increased in the frontal recess of eosinophilic chronic rhinosinusitis patients. J Otolaryngol Head Neck Surg. (2017) 46(1):36. doi: 10.1186/s40463-017-0214-2

PubMed Abstract | Crossref Full Text | Google Scholar

43. Xuan L, Zhang N, Wang X, Zhang L, Bachert C. IL-10 family cytokines in chronic rhinosinusitis with nasal polyps: from experiments to the clinic. Front Immunol. (2022) 13:947983. doi: 10.3389/fimmu.2022.947983

PubMed Abstract | Crossref Full Text | Google Scholar