Identification of host-microbiome interactions in nasal diseases using multiomics integration

doi:10.21203/rs.3.rs-4962429/v1

Download PDF

Research Article

Identification of host-microbiome interactions in nasal diseases using multiomics integration

https://doi.org/10.21203/rs.3.rs-4962429/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

An imbalance in the nasal microbiome is thought to be closely related to the development of nasal diseases. However, nasal microbiome-host interactions have rarely been studied.

Objective

The aim of this study was to comprehensively investigate the cross-talk between mucosal gene expression and the microbiota in patients with chronic rhinosinusitis with nasal polyps (CRSwNP) and nasal inverted papilloma (NIP).

Methods

We performed a cross-sectional study of 43 patients with CRSwNP, 27 patients with NIP and 34 controls using 5R 16S rRNA gene sequencing. A total of 40 CRSwNP samples, 20 NIP samples and 20 control samples were analyzed according to host transcriptome.

Results

This study describes the microbiome characteristics of the specific nasal mucosal microenvironment in patients with CRSwNP and NIP. In CRSwNP and NIP samples, host gene-bacteria interaction analysis revealed multiple host pathways that were associated with the nasal microbiota, mainly including multiple host pathways such as those related to immunity, metabolism, host defense, and cell proliferation. In addition, in both nasal diseases, the shared host pathways that were associated with the nasal microbiota were mainly immune response-related pathways, such as the NF-kappa B signaling pathway. In CRSwNP, disease-specific pathways that were associated with the nasal microbiota were mainly related to host recognition and the immune response, while in NIP, disease-specific pathways were mainly related to cell proliferation. Based on Bayesian network analysis, we found that the abundance of Geobacillus stearothermophilus in nasal polyps was significantly correlated with the NF-kB pathway activation, and we further proved this correlation.

Conclusion

Our study highlights the complex interplay between the nasal microbiota and host-population patterns, with disease-specific and shared host gene-microbiome associations associated with different features of nasal disease. Our findings may provide new insights into the pathophysiology of nasal diseases and a theoretical basis for future microbiota-based treatment strategies for nasal diseases.

chronic rhinosinusitis with nasal polyps

nasal inverted papilloma

nasal microbiome

microbiome-host interaction

Nasal diseases affect hundreds of millions of people worldwide and impose heavy health and economic burdens on society. Currently, the treatment of nasal diseases remains comprehensive, and it is based on surgery and supplemented with drugs. However, because the cause is still unclear, some patients still experience repeated attacks, seriously affecting the quality of life of patients.^1–4 Therefore, it is of great significance to explore the etiology for understanding the physiologic process of nasal diseases and seeking new treatment methods.

There are rich microbial communities in the nasal cavity, and there are complex interactions between these microbial communities and the host.^5–6 Previous studies have suggested that pathogenic microorganisms, including Staphylococcus aureus and human papillomavirus, may play important roles in the pathophysiology of nasal diseases.^7–8 However, these pathogens account for the cause of disease in only a small number of patients, and there are still a large number of patients for whom the cause of disease is unknown. In recent years, many studies have suggested that a local bacterial imbalance occurs in the nasal cavity of patients with nasal diseases, but the effect of this bacterial imbalance on the host has not been reported.^9–10 To this end, by simultaneously analyzing the microbiota and host transcriptome, we initially explored whether other members of the microbiota in addition to S. aureus play important roles in the pathophysiological processes of nasal diseases.

Microbial-host interactions may be at the core of nasal mucosal homeostasis, and microbial imbalance may cause or exacerbate disease. Therefore, the accessibility of the microbiome and diseased tissues provides a unique opportunity to study host-microbiome interactions in the pathophysiological processes of nasal disease.¹¹ To gain insight into these host-microbe interactions in the nasal mucosa, we comprehensively analyzed the relationship between mucosal gene expression and the microbiome in patients with chronic sinusitis with nasal polyps (CRSwNP) and nasal inverted papilloma (NIP) (Fig. 1). CRSwNP and NIP represent nasal inflammation and tumors, respectively, providing the opportunity to compare the two nasal diseases and the associated microbiomes and transcriptomes. We demonstrate for the first time that these two diseases have unique mucosal microbiota signatures. In addition, for the first time, we elucidated the associations between nasal microbiota and host genes and further characterized disease-specific and shared host gene-microbiome associations for both diseases. These findings may provide new insights into the pathophysiology of nasal diseases and a theoretical basis for future microbiota-based treatment strategies for nasal diseases.

Subject recruitment and sampling

In this study, a total of 43 patients with bilateral CRSwNP, 27 patients with NIP and 34 controls were recruited. All of these subjects underwent surgery, and tissue was retained during surgery for subsequent sequencing analysis. Patients with CRSwNP met the diagnostic criteria of the EPOS2020 guidelines.¹ NIP was determined by postoperative histopathology. The turbinate mucosa tissues of patients with a deviated nasal septum and cranial base operation were included in the control group. The study was approved by the Ethics Committee of Tianjin First Central Hospital. All the subjects were informed in advance and signed informed consent forms.

Each patient underwent a physical examination by two rhinologists to confirm the diagnosis, and clinical information such as age, sex, smoking history, medication history, and surgical history was recorded.

The exclusion criteria were as follows: ① Patients with immune-related diseases, genetic disorders, pregnancy, clotting disorders, or cystic fibrosis. ② Patients with unilateral nasal polyps or infection or anatomic sinusitis. ③ Patients who received systemic antibiotics within 12 weeks prior to screening. ④ Patients who received systemic immunosuppressive therapy within 12 weeks prior to screening. ⑤ Patients who received systemic or local steroid hormones within 12 weeks prior to a screening history of relevant surgery.

All the patients underwent surgery. After anesthesia, nose hairs were trimmed, and the skin of the nose and nasal vestibule was disinfected to reduce contamination. Under the guidance of nasal endoscopy, polyp tissue, tumor tissue and normal turbinate mucosa were collected in sterile tubes, and the contaminants on the surface of the retained tissue were washed away with phosphate-buffered saline (PBS). The samples were stored in RNAlater (Sigma-Aldrich) in sterile empty tubes and immediately stored in a refrigerator for subsequent detection and analysis. In addition, simulated samples containing PBS alone were collected at each sampling to assess environmental contamination. Preliminary experiments were performed to assess and confirm the quality and uniformity of the collected samples, as well as the feasibility of the method, before starting a large sample collection. All the procedures were performed in the operating room under sterile conditions.

5R 16S rRNA gene sequencing and analyses

To better characterize the microbiota in nasal tissues, 5R 16S sequencing was used in this study.¹² In brief, nasal tissue and negative controls were extracted via the Cetyl Trimethyl Ammonium Bromide method. All negative controls included sampling controls, DNA extraction controls and no-template PCR amplification controls. The modified 5R 16S rRNA gene was amplified. The modified 5R 16S rRNA gene is composed of five regions (the V2, V3, V5, V6, and V8 regions). The amplified products were then purified and quantified by standardized means. The purified products were then sequenced on the Illumina NovaSeq platform supported by Lc-Bio Technologies Co., Ltd. (Hangzhou, China). After the sample sequencing data were removed, the high-quality sequences that remained after the removal of low-quality sequencing results were used for subsequent analysis.

The reads were demultiplexed per sample, filtered and aligned to each of the five amplified regions based on the primer sequences. The SMURF (short multiple regions framework) method was applied to combine read counts from the five regions into coherent profiling results to solve the maximum likelihood problem.¹³ Then, the taxonomic identification and relative abundance calculation of bacteria were carried out. The database that was used in this project is optimized Greengenes (May 2013 version).

Subsequent microbiome diversity analysis and microbiome difference analysis were carried out on the basis of these data. The richness and uniformity of alpha diversity are mainly reflected by indices such as Chao1 and observed species. Beta diversity analysis usually begins by calculating the distance matrix between environmental samples, which includes the distance between any two samples. Beta diversity, together with alpha diversity, constitutes the overall diversity or the biological heterogeneity of a given environmental community. The Kruskal-Wallis test was used to compare multiple groups with biological duplicate samples.

5R 16S rRNA gene sequencing and analyses of negative controls

To reduce the impact of low-abundance noise on subsequent analysis, the sequence read number of each sample was normalized, and samples with total reads < 1000 (including negative controls) and bacterial data with a relative abundance < 10^− 4 were removed. Then, according to the prevalence of bacteria in the negative control samples, we determined which bacteria were the contaminating bacteria at the sampling end and the experimental end. The principle and process of impurity removal were as follows. Five negative control samples were sequenced. The genus-level species that were present in more than 50% of the samples (more than 3 samples and more than 0.001% of the abundance) according to the sequencing results were identified as contaminating bacteria (or heterobacteria). Contaminating bacteria were removed from the nasal tissue sequencing results. The removed species were normalized again to obtain the relative abundance of real species in the sample.

Bioinformatics analysis of RNA-seq data

As previously mentioned¹⁴, the RNA was isolated and purified from the total sample using a standardized process. Then, the quantity and purity of the isolated RNA were controlled, and the integrity of the RNA was tested. The concentration was > 50 ng/µL, the RNA integrity number (RIN) was > 7.0, and the total RNA amount was > 1 µg. After two rounds of purification, mRNA with poly(A) (polyadenylate) was specifically captured. The captured mRNA was fragmented at high temperature with the NEBNextR RNA Fragmentation Module (NEB, cat. e6150, USA) at 94°C for 5–7 minutes. cDNA was synthesized from the fragmented RNA with Invitrogen SuperScriptTM II Reverse Transcriptase (Invitrogen, cat. 1896649, USA). E. coli DNA polymerase I (NEB, cat.m0209, USA) and RNase H (NEB, cat.m0297, USA) were then used for two-strand synthesis. To preserve the orientation information of the transcript during transcriptome sequencing, these complex double strands of DNA and RNA were converted into double-stranded DNA, and dUTP Solution (Thermo Fisher, cat. R0133, CA, USA) was added to the double strands to convert the ends of the double-stranded DNA into flat ends. Then, an A base is added to each end so that it could be connected to the terminal with T base joints, and the size of the fragment was screened and purified by magnetic beads. The double-stranded library was digested with UDG enzyme (NEB, cat. m0280, MA, US) and then formed by PCR with a fragment size of 300 bp ± 50 bp (chain-specific library). Finally, we used an Illumina NovaSeqTM 6000 (LC Bio Technology Co., Ltd.; Hangzhou, China) to perform double-end sequencing in PE150 mode according to standard procedures. The sequence information of the transcripts that were obtained by sequencing was only derived from the first strand.

After using Cutadapt to removed unqualified sequences (sequencing joints, low-quality sequences, etc.) from the original data to obtain valid data (clean data), reference genome alignment was performed using HISAT2. Based on the HISAT2 comparison results, Stringtie was used to reconstruct the transcripts and calculate the expression levels of all the genes in each sample. Gene expression level analysis mainly aimed to analyze protein-coding genes (mRNAs) that were annotated by the genome, and the expression levels of genes were statistically measured to evaluate the correlation of gene expression characteristics and differentially expressed genes within and between groups. When measuring gene expression levels, fragments per kilobase million (FPKM) values, which are standardized based on the original read counts of genes, were used as measures of gene expression levels, and gene expression levels in different samples were quantified. In this project, a fold change > = 2 (that is, the absolute value of log2FC > = 1) was considered the change threshold, and a q value < 0.05 (the q value is the correction value of the p value) was considered the criterion for identifying differentially expressed genes (|log2FC|>= 1&q < 0.05). The results of differential expression gene analysis, differential expression gene enrichment analysis and GSVA were obtained in the set comparison group.

Preprocessed microbiome and host transcriptome data

Based on the approach of Priya et al.,¹¹ the following process was performed on microbiome data for each disease cohort separately. First, sequences of archaea, chloroplasts, known contaminants from laboratory reagents or kits, and environmental contaminants that were associated with soil or water were removed from the microbial abundance table. Next, microbial abundance tables were constructed at the phylum, class, order, family, genus, and species levels and filtered based on abundance and prevalence, retaining only taxa that had relative abundances above 0.001 in at least 20% of the samples. The microbiome data of each disease cohort were independently processed to obtain the taxon abundance matrix of each disease cohort, which included 134 taxa in CRSwNP group and 156 taxa in NIP group.

In host transcription data processing, genes with low expression were first removed to ensure that each gene was expressed in at least half of the samples in the disease cohort. We used the R package "DESeq2" (version 1.22.2) for variance stabilizing transformation of the filtered gene expression read count and to filter the 25% of genes that had the least variance. RNA-seq data from each disease cohort were independently subjected to these steps, producing a unique host gene expression matrix for the disease for downstream analysis, including 23,030 genes in CRSwNP group and 22,849 genes in NIP group.

Sparse CCA analysis

We used a machine learning framework that was developed by Priya et al. to integrate high-dimensional datasets of host gene expression and microbiome abundance.¹¹ SparseCCA was used to identify host genes that are associated with microorganisms to characterize pathway-level associations. The analytical framework was applied to paired host gene expression data and microbiome data for each disease cohort and control group, respectively, to avoid potential batch effects. For each disease cohort dataset, we considered only associations that were observed in patients not in controls. Prior to the application of this analytical framework, host gene expression matrix and microbiome abundance matrix data were standardized and normalized to meet the distribution requirements of statistical models.

SparseCCA was used to determine group-level correlations between paired host gene expression and microbiome data in each disease cohort. sparseCCA performs feature selection via the L1 or lasso penalty while maximizing the correlation between the two datasets. Its objective function can be expressed as:

$$\:{maximize}_{u,v}{u}^{T}{X}^{T}Yv\:subject\:to\:{u}^{T}{X}^{T}Xu\le\:1,{v}^{T}{Y}^{T}Yv\le\:1,‖u‖{}_{1}\le\:{{\lambda\:}}_{1}，,‖v‖{}_{1}\le\:{{\lambda\:}}_{1}，$$

where X and Y represent two data matrices with the same number of samples but different numbers of features (representing microbiome taxa composition data and host gene expression data, respectively). u and v are the canonical load vectors of X and Y, respectively; λ1 and λ2 control the lasso penalty of U and V, respectively. τ represents the transpose of a matrix.

As with the original method, leave-one-out cross-validation was used for a grid-search approach to obtain the optimal hyperparameters. Using this method, λ1 and λ2 for group A were set to 0.2 and 0.15, respectively, and λ1 and λ2 for group B were set to 0.266 and 0.177, respectively. After the sparsity parameters were determined, the sparse CCA model was fitted to obtain a subset of the relevant host genes and microorganisms (called components), calculating only the top 10 components for each disease cohort. Additionally, the leave-one-out cross-validation approach was used to calculate the importance of each component. Cor.test was used to evaluate the true strength and significance of the association, and Benjamini‒Hochberg (FDR) was used to correct the P value of the multiple hypothesis test for each disease cohort. Only significant components with FDR < 0.1 were retained. Based on this method, three important components in group A were identified, with an average of 1739 host genes and 4.5 microorganisms. There were six significant components in group B, with an average of 2,786 host genes and 6.4 microbes. sparseCCA was applied separately for each disease cohort using the R language (version 4.2.0) package "PMA" (version 1.2.1). All the components were visualized using Cytoscape (version 3.10.0).

Inference of microbial-host interaction networks

In addition to identifying differential pathways, identifying important directional edges between pathways can also provide valuable insights when studying microbe-host interactions. The path interaction network (Bayesian Network analysis) was used with the "CBNplot" package to construct the gene interaction network¹⁵. Based on the host expression profile data, associated microorganisms were screened out in combination with sparseCCA, the interaction direction was calculated, the bnpathplot function was used to obtain the interaction direction and intensity, and Cytoscape was used to plot the network.

Fluorescence in situ hybridization (FISH)

FISH is performed using probes that target the 16S rRNA gene sequence for a specific bacterial taxon. According to previous methods¹⁶, the Geobacillus FISH probe was hybridized to tissue sections and labeled with FITC at the 5' and 3' ends (5’CCGAATCAAGGCAAGCCCCAATC-3’). This probe was designed and synthesized by Exonbio (Gungzhou, China). This probe targets Geobacillus stearothermophilus, and some Geobacillus sp. EUB330 proteins target a conserved domain of bacterial 16S rRNA. FISH images were captured with a Nikon 80i microscope. The images were analyzed and scored according to the fluorescence signal.

Western blotting

Tissues were homogenized in liquid nitrogen, and an appropriate amount of RIPA cell lysis was added. After centrifugation, the concentration of superalbumin was extracted and determined. The protein concentration was determined by the BCA method. The proteins were denatured by boiling after the original volume of buffer was added. After protein electrophoresis and membrane transfer, the membranes were blocked in BSA solution at room temperature, and the membranes were incubated with primary antibodies (rabbit anti-p65, ab32536; rabbit anti-p-p65, ab109458; rabbit anti-β-actin, ab8227) at 4°C overnight. The next day, after the membrane was washed with the washing liquid and incubated at room temperature with the secondary antibody for 1 hour. After the membrane was washed with the washing liquid, ECL was applied for color development. The original gels were showed in Supplymentary Fig. 1.

The nasal microbiomes of CRSwNP and NIP patients are distinct

In this study, 43 patients with bilateral CRS, 27 patients with NIP, and 34 controls were eventually enrolled. The demographic data of the patients are described in Supplementary Table 1. To better identify the microbiome in the tissues, we used 5R 16S rDNA sequencing to sequence the microbiome in the nasal tissues. With increasing sequencing depth, the sparse curves at the species level tended to stabilize, indicating that 5R 16S rDNA coverage was sufficient, and the sequencing results could stably represent the species information of the sample (Supplymentary Fig. 2A and B). To assess the abundance and uniformity of the sample species composition within the different groups, the α diversity of the different groups was compared. The α diversity was lower (Supplymentary Fig. 2C) in the CRSwNP group than in the control group. There was no significant difference in the α diversity between the NIP group and the control group (Supplymentary Fig. 2D).

There were partial but significant differences in classification characteristics between patients with CRSwNP and controls (R² = 0.019, P<0.001; Fig. 2A). Similar findings were observed in NIP patients and controls (R² = 0.047, P = 0.045; Fig. 2B). In addition, a partial but significant difference in categorical characteristics was observed between patients with CRSwNP and those with NIP (R² = 0.250, P<0.001; Supplymentary Fig. 2E). These results suggest that patients with nasal diseases have different microbiome characteristics from those of normal people and that the two diseases have their own unique microbiome characteristics.

At the phylum and genus levels, the three groups exhibited different microbiome compositions. At the phylum level (Supplymentary Fig. 2F and G), Proteobacteria, Actinobacteria and Firmicutes were the most abundant phyla in the CRSwNP and NIP groups. At the genus level (Fig. 2C and D, Supplymentary Fig. 3A and B), Aquabacterium was the most abundant species in patients with CRSwNP, while Corynebacterium was the most abundant species in patients with NIP. To identify differentially abundant taxa, LEfSe analysis of the nasal microbiota composition was performed for 43 patients with bilateral CRS, 27 patients with nasal varus papilloma, and 34 controls (Fig. 2E and F). The results suggested that there were significant differences in relative abundance between the two groups (LDA score>2.0, p<0.05). At the genus level, 53 species were differentially abundant between the CRSwNP group and the control group according to the read classification. The relative abundances of Aquabacterium and Sphingomonas increased to the greatest extents in CRSwNP patients and the abundances of Bacteroides were decreased to the greatest extents (Supplymentary Fig. 3C). The 14 species based on the read classification were difference-rich. The relative abundances of Hemophilus and Mycobacterium were increased to the greatest extend in NIP patients and that of Gardnerella was decreased to the greatest extent (Supplymentary Fig. 3D).

Host transcriptomes of CRSwNP and NIP patients

The host transcriptomes of 40 CRSwNP samples, 20 NIP samples, and 20 control samples were analyzed. Principal component analysis (PCA) revealed significant differences among the CRSwNP, NIP, and control groups (Fig. 3A, Supplymentary Fig. 4A, B and C). To identify potential target genes that were closely related to CRSwNP and NIP, differential expression analysis of the CRSwNP, NIP and control groups was performed (FDR < 0.01, fold change (FC) > 1.5). A total of 4456 differentially expressed genes were identified between the CRSwNP group and the control group (Fig. 3B). KEGG analysis revealed that the top 20 pathways associated with differentially expressed gene enrichment were cytokine − cytokine receptor interaction, neuroactive ligand − receptor interaction and metabolic pathways and pathways in cancer (Fig. 3D). There 6040 differentially expressed genes between the NIP and control groups (Fig. 3C). KEGG analysis suggested that the top 20 pathways of differentially expressed gene enrichment included metabolic pathways, pathways in cancer, neuroactive ligand-receptor interactions, cytokine-cytokine receptor interactions and other pathways (Fig. 3E). These results suggest that the immune, metabolic and cell proliferation pathways of the CRS and NIP groups are significantly different from those o0f the control group. Therefore, our further analysis suggested that different groups had different differentially expressed genes involved in immune, metabolic and cell proliferation pathways (Fig. 3F).

The disease-specific nasal microbiome is associated with different host pathways

Since different nasal diseases have unique nasal microbiota characteristics, we hypothesized that changes in the CRSwNP and NIP transcriptomes may be partly related to the nasal microbiota and that there may also be microbiota-host interactions involved in nasal diseases. To investigate this possibility, we performed transcriptomic sequencing of the samples at the same time as microbial sequencing. Transcriptome sequencing was performed on the remaining nasal tissues from 104 patients according to strict NGS RNA-seq criteria, and ultimately, we obtained 80 pairs of paired data related to the nasal microbiome and host gene expression, including 40 pairs in the nasal polyp cohort, 20 pairs in the NIP cohort, and 20 pairs in the control cohort. Follow-up analysis was also conducted.

Previous studies have suggested that microbes that are involved in the same biological function may interact with host genes in the form of groups. Considering the challenges of integrating multiomics data with high dimensionality, sparsity, and multicollinearity, we used sparse CCA to explore nasal microbiome‒host interactions. This approach facilitates characterization of the association between host transcriptome expression and nasal microbiome abundance in both nasal diseases at the population level. In this study, sparseCCA was first used to reduce the dimension of the gene expression profile and microbial abundance table to gather potentially related features together and then to identify multiple microbial and host pathways/genes that may be related to reduce the computational burden and interference features of the next analysis and improve the analysis accuracy. SparseCCA is used to define host genes or microorganisms with potentially related features that clustered together as components. By fitting sparse CCA models to the transcriptome and microbiome datasets of each disease cohort, host gene components that were significantly associated with the nasal microbiome were identified. We then performed pathway enrichment analyses of host genes for these components, which were significantly associated with the nasal microbiome, to identify the host pathways that were associated with the nasal microbiome in both nasal diseases.

In this study, we found a population-level association between host transcriptome expression and nasal microbiome abundance in two nasal diseases (Supplementary Tables 2 and 3). In the nasal polyp group, there was a correlation between the host transcriptome and the abundance of the nasal microbiome in three components. In the NIP group, associations between host transcriptome expression and nasal microbiome abundance were identified in six components. Through preliminary gene annotation and pathway analysis, 226 host pathways were identified in the two nasal diseases, mainly including immune, metabolic, host defense and cell proliferation pathways (Supplementary Table 4). These results suggest that the host transcription profile of nasal diseases patients may be partially influenced by the nasal microbiota.

To visualize the host transcription profiles that are affected by nasal microbiota in nasal diseases, we performed GSVA on the enriched pathways in the associated components and identified the top 20 pathways for each component (Fig. 4A, Supplementary Table 5). We found that nasal microbes mainly affect host metabolism, the immune response, host defense function, and other processes. In addition, we found that the effects of the nasal microbiome on host transcriptional profiles are significantly different in different nasal diseases. Among them, in CRSwNP, the nasal microbiome mainly affects host metabolism-related and immune response-related pathways. For example, Th17 cell differentiation, Th1 and Th2 cell differentiation and the NF-kappa B signaling pathway have been proven to have important effects on the pathophysiological processes of CRSwNP. In NIP, nasal bacteria affect signal transduction pathways. For example, glycosaminoglycan biosynthesis-keratan sulfate is closely related to the occurrence and development of many tumors. In addition, it affects a variety of amino acid metabolic pathways, such as D-glutamine and D-glutamate metabolism. The nasal microbiome may play a crucial role in the occurrence and development of different nasal diseases.

In addition, previous studies have shown that there are "shared" and "disease-specific" pathways in the intestinal microbiome in different intestinal diseases. The "shared" pathway is the host pathway that is associated with the microbiome that is present in different diseases. Additionally, "disease-specific" pathways are host pathways that are associated with the microbiome that are present only in a single disease. Therefore, we hypothesized that this “shared” and “disease-specific” pathway also exists in nasal diseases. In our study, we found a total of 89 shared that are pathways associated with the nasal microbiome (Supplementary Table 4). For simplicity, we focused on the top ten most important shared and disease-specific pathways (Fig. 4B). We found that the shared pathways that are associated with nasal disease in both nasal disease groups were associated with immune response-related pathways. Examples include Th17 cell differentiation, Th1 and Th2 cell differentiation, and the NF-kappa B signaling pathway (Fig. 4C). These pathways are associated with local immune disorders and the promotion of inflammation in sinusitis. In NIP, Th17 cell differentiation and Th1 and Th2 cell differentiation are associated with the breakdown of the epithelial barrier. These results suggest that the nasal microbiome can promote the development of disease by promoting the immune response in patients with different nasal diseases.

In addition, we identified "disease-specific" pathways that are associated with nasal flora, including 24 CRSwNP-specific pathways and 113 NIP-specific pathways (Supplementary Table 4). In CRSwNP, the nasal microbiome mainly affects host recognition and the immune response. For example, the B cell receptor signaling pathway plays a crucial role in the effect of various immune cells in CRSwNP, and the nasal microbiome plays a role through the bacterial invasion of epithelial cells pathway (Fig. 4D). The nasal microbiome in NIP mainly affects host functions. For example, the mTOR signaling pathway can facilitate this process (Fig. 4D). These results suggest that the nasal microbiome plays a distinct and important role in different nasal diseases and may play a crucial role in the occurrence and development of the disease.

The abundance of G. stearothermophilus in CRS is correlated with NF-kB pathway activation

These results suggest that the host transcription profile of nasal diseases may be partially influenced by the nasal microbiota. To provide an initial validation of our findings, based on Bayesian networks, we looked for the strongest possible pair of bacteria-host pathway relationships among the different components. Based on Bayesian networks, we found that G. stearothermophilus in component 3 of the CRSwNP group could significantly affect the host NF-kB pathway (Fig. 5A). Spearman correlation analysis further revealed that the abundance of G. stearothermophilus was positively correlated with the NF-kB pathway (Fig. 5B and C). To verify this finding, FISH was used to verify the presence of G. stearothermophilus in CRS samples (Fig. 5D). In addition, based on the expression of high- and low-fluorescence density of G. stearothermophilus, the samples were divided into high- and low-fluorescence density of G. stearothermophilus groups. WB results showed that the gray value of p-P65 increased in high-fluorescence density of G. stearothermophilus in CRSwNP, but the protein expression of P65 was not affected (Fig. 5E and F). Our results suggest that higher expression of NF-kB pathway activation was siginificantly observed in high-fluorescence density group than in low-fluorescence density group in nasal polyps. Based on a series of analyses, such as sparseCCA and Bayesian network analyses, we screened the effect of G. stearothermophilus on the host NF-kB pathway and verified its interaction through immunofluorescence experiments. Our analytical method can be used to screen for more reliable and accurate microbe‒host interaction mechanisms from large amounts of microbial and transcriptome data within a limited computational link, improve the analysis efficiency, and provide a new approach for the study of nasal diseases.

There are rich microbial communities in the nasal cavity, and there are complex interactions between these communities and the host.^17–18 Previous studies have described an imbalance in the nasal microbiota, but the relationships among the microbiota, the host, and disease pathogenesis have been poorly studied.^5–6,9−10 To gain insight into these host–microbe interactions in the nasal mucosa, we comprehensively analyzed the relationship between mucosal gene expression and the microbiome in patients with CRSwNP and NIP. We demonstrate for the first time that these two diseases have unique mucosal microbiota signatures. In addition, for the first time, we elucidated the associations between nasal microbiota and host genes and further characterized disease-specific and shared host gene‒microbiome associations for both diseases. These findings may provide new insights into the pathophysiology of nasal diseases and a theoretical basis for future microbiota-based treatment strategies for nasal diseases.

The mucosa is a dynamic interface between host and microbial ecosystem networks, and in recent years, multiple studies have suggested that the mucosal microbiota plays a crucial role in the progression of multiple chronic diseases. For example, the mucosal microbiome is closely related to host immune function in autoimmune diseases.¹⁴ There is a correlation between the abundance of pathogenic mucosal bacteria in colon cancer and the expression of host genes that are associated with gastrointestinal inflammation and tumorigenesis.^19–20 In this study, we demonstrated that mucosae from patients with different nasal diseases have their own microbiota characteristics. Although multiple factors may contribute to nasal disease, our systematic analysis highlights the potential role of microbial communities in the development of nasal disease. This is significant because many previous studies on the nasal microbiome have been based on nasal swabs, and this nasal microbiome imbalance may reflect disease status but not the disease mucosal microenvironment. Therefore, the unbalanced bacterial flora found in the nasal cavity in previous studies may be a passenger rather than a driver in the development of nasal disease. For the first time, we focused on changes in the mucosal microflora of the nasal cavity, which are more likely to reflect the mucosal microenvironment of different nasal diseases. Our results suggest that a considerable degree of dysbiosis may occur in the nasal environment of patients with nasal disease. Future studies using mouse models of their potential pathogenic functions will reveal whether these candidates are drivers or passengers in the development of nasal diseases.

Previous studies have only described changes in the microbiome of nasal disease patients, and the impact of these changes on the host is unknown. To this end, we further characterized the association between the nasal microbiome and host genes by analyzing both the microbiome and the host transcriptome in nasal tissue for the first time. In this regard, through biogenic analysis, we found that nasal microorganisms may affect the host immune, metabolic, defense and cell proliferation pathways. Some of these bacteria-host interaction pairs have been confirmed. For example, Lactobacillaceae can improve diabetes-related symptoms by activating the PI3K-Akt signaling pathway in patients with diabetes. Additionally, Lactobacillus was found to be closely related to propanoate metabolism in Parkinson's disease patients^21–22. These results suggest that the host transcription profile of patients with nasal disease may be partially influenced by the nasal microbiota. In addition, based on a series of analyses, such as sparseCCA and Bayesian network analyses, we determined that the abundance of Geobacillus stearothermophilus in nasal polyps was significantly correlated with the activation of the NF-kB pathway in the host. This correlation was confirmed by immunofluorescence experiments. Our analytical method can be used to screen for more reliable and accurate microbe‒host interaction mechanisms from large amounts of microbial and transcriptome data within a limited computational link, improve the analysis efficiency, and provide a new approach for the study of nasal diseases.

Previous studies have focused only on the pathogenic role of microflora in a single disease, and there is no research on whether microbiome play the same or different roles in different diseases. Here, we analyzed the relationship between mucosal gene expression and the microbiome in patients with different nasal diseases. We found that the shared pathways that were associated with nasal disease in both nasal disease groups were associated with immune response-related pathways. Examples include Th17 cell differentiation, Th1 and Th2 cell differentiation, and NF-kappa B signaling pathways.^23–25 These results suggest that the nasal flora can promote the development of disease by promoting the immune response in patients with different nasal diseases. In addition, we also identified "disease-specific" pathways that are related to the nasal flora, such as the B cell receptor signaling pathway, which plays a crucial role in the effects of multiple immune cells in CRSwNP.^26–27 The nasal flora in patients with NIP mainly affects host functions. For example, the mTOR signaling pathway can facilitate this process.²⁸ These results suggest that the nasal microbiome plays a distinct and important role in different nasal diseases and may play a crucial role in the occurrence and development of disease.

This study presents a comprehensive landscape of nasal mucosal host–microbe interactions in two nasal diseases. Our study is the first to characterize the microflora in the specific nasal mucosal microenvironment of patients with CRSwNP and NIP. Most importantly, for the first time, we illustrate the complex interactions between nasal microbiota and host-population patterns with disease-specific and shared host gene-microbiome associations that are associated with different nasal disease features. These findings can guide the development of future studies on mechanisms underlying gene-bacteria interactions and serve as a resource for the rational selection of therapeutic targets for rhinopathy. Our findings may provide new insights into the pathophysiology of nasal diseases and a theoretical basis for future microbiota-based strategies for treating nasal diseases. In addition, our analytical method can identify more reliable and accurate microbe‒host interaction mechanisms from large amounts of microbial and transcriptome data within limited computational links, improve analysis efficiency, and provide new ideas for the study of nasal diseases.

Ethical approval and consent to participate
This study was approved by the Ethics Research Committees of the Tianjin First Central Hospital (Approval number. 2021N037KY). This study was performed in accordance with the relevant guidelines and regulations. Written informed consent was obtained from all participants.

Consent for publication
Not applicable

Competing interests

The authors declare that they have no competing interests.

Funding

The authors thank all the subjects who participated in this study. This work was supported by Tianjin Health Research Project (TJWJ2022XK020); National Natural Science Foundation of China (82401333); and Tianjin Natural Science Foundation (19JCYBJC27200). This work was funded by Tianjin Key Medical Discipline Construction Project.

Author Contribution

YL, ZC, JZ, and GZ all developed the study concept and design. CZ, ZL and JL collected nasal samples. CZ, WS and JL guided statistical analysis and data analysis. YL, ZC, JZ, and GZ verified the experimental design, visualized the experimental results, and critically reviewed the manuscript. YL, CZ, ZL and JL recruited patients and collected specimens, collected clinical metadata, interpreted the microbiome data, and were major contributors in writing the manuscript and reviewing it critically. The authors read and approved the final manuscript.

Acknowledgement

The authors thank all the subjects who participated in this study. This work was funded by Tianjin Key Medical Discipline Construction Project.

Data Availability

We declare that the main data supporting the finding of this study were available within the paper and its Supplementary Information. The clean reads were deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) database (Accession no. PRJNA997619). The raw transcriptomic data for the human cohort have been deposited in the NCBI under GSE255573 for controlled access.

Fokkens WJ, Lund VJ, Hopkins C, Hellings PW, Kern R, Reitsma S, et al. European Position Paper on Rhinosinusitis and Nasal Polyps 2020. Rhinology. 2020 Feb 20;58(Suppl S29):1-464.
Fokkens WJ, Viskens AS, Backer V, Conti D, De Corso E, Gevaert P, et al. EPOS/EUFOREA update on indication and evaluation of Biologics in Chronic Rhinosinusitis with Nasal Polyps 2023. Rhinology. 2023 Jun 1;61(3):194–202.
Yu S, Grose E, Lee DJ, Wu V, Pellarin M, Lee JM. Evaluation of inverted papilloma recurrence rates and factors associated recurrence after endoscopic surgical resection: a retrospective review. J Otolaryngol Head Neck Surg. 2023 Apr 27;52(1):34.
Lee JJ, Roland LT, Licata JJ, Orlowski HLP, Jiramongkolchai P, Piccirillo JF, et al. Morphologic, intraoperative, and histologic risk factors for sinonasal inverted papilloma recurrence. Laryngoscope. 2020 Mar;130(3):590–596.
Liang Y, Xie R, Xiong X, Hu Z, Mao X, Wang X, et al. Alterations of nasal microbiome in eosinophilic chronic rhinosinusitis. J Allergy Clin Immunol. 2023 May;151(5):1286–1295.e2.
Zhao Y, Chen J, Hao Y, Wang B, Wang Y, Liu Q, et al. Predicting the recurrence of chronic rhinosinusitis with nasal polyps using nasal microbiota. Allergy. 2022 Feb;77(2):540–549.
Du K, Zhao Y, Zhang X, Li C, Hao Y, Du X, et al. Staphylococcus aureus lysate induces an IgE response via memory B cells in nasal polyps. J Allergy Clin Immunol. 2024 Mar;153(3):718–731.e11.
Rha MS, Kim CH, Yoon JH, Cho HJ. Association of the human papillomavirus infection with the recurrence of sinonasal inverted papilloma: a systematic review and meta-analysis. Rhinology. 2022 Feb 1;60(1):2–10.
Kim JH, Kim SH, Lim JY, Kim D, Jeong IS, Lee DK, et al. Association between the sinus microbiota with eosinophilic inflammation and prognosis in chronic rhinosinusitis with nasal polyps. Exp Mol Med. 2020; 52(6):978–987.
Kidoguchi M, Imoto Y, Noguchi E, Nakamura T, Morii W, Adachi N, et al. Middle meatus microbiome in patients with eosinophilic chronic rhinosinusitis in a Japanese population. J Allergy Clin Immunol. 2023 Dec;152(6):1669–1676.e3.
Priya S, Burns MB, Ward T, Mars RAT, Adamowicz B, Lock EF, et al. Identification of shared and disease-specific host gene-microbiome associations across human diseases using multi-omic integration. Nat Microbiol. 2022 Jun;7(6):780–795.
Nejman D, Livyatan I, Fuks G, Gavert N, Zwang Y, Geller LT, et al. The human tumor microbiome is composed of tumor type-specific intracellular bacteria. Science. 2020 May 29;368(6494):973–980.
Fuks G, Elgart M, Amir A, Zeisel A, Turnbaugh PJ, Soen Y, et al. Combining 16S rRNA gene variable regions enables high-resolution microbial community profiling. Microbiome. 2018 Jan 26;6(1):17.
Hu S, Bourgonje AR, Gacesa R, Jansen BH, Björk JR, Bangma A, et al. Mucosal host-microbe interactions associate with clinical phenotypes in inflammatory bowel disease. Nat Commun. 2024 Feb 17;15(1):1470.
Sato N, Tamada Y, Yu G, Okuno Y. CBNplot: Bayesian network plots for enrichment analysis. Bioinformatics. 2022 May 13;38(10):2959–2960.
Yu J, Chen Y, Fu X, Zhou X, Peng Y, Shi L, et al. Invasive Fusobacterium nucleatum may play a role in the carcinogenesis of proximal colon cancer through the serrated neoplasia pathway. Int J Cancer. 2016 Sep 15;139(6):1318-26. doi: 10.1002/ijc.30168. Epub 2016 May 17.
Sasso JM, Ammar RM, Tenchov R, Lemmel S, Kelber O, Grieswelle M, et al. Gut Microbiome-Brain Alliance: A Landscape View into Mental and Gastrointestinal Health and Disorders. ACS Chem Neurosci. 2023 May 17;14(10):1717–1763.
Lin L, Yi X, Liu H, Meng R, Li S, Liu X, et al. The airway microbiome mediates the interaction between environmental exposure and respiratory health in humans. Nat Med. 2023 Jul;29(7):1750–1759.
Ralser A, Dietl A, Jarosch S, Engelsberger V, Wanisch A, Janssen KP, et al. Helicobacter pylori promotes colorectal carcinogenesis by deregulating intestinal immunity and inducing a mucus-degrading microbiota signature. Gut. 2023 Jul;72(7):1258–1270.
Fu K, Cheung AHK, Wong CC, Liu W, Zhou Y, Wang F, et al. Streptococcus anginosus promotes gastric inflammation, atrophy, and tumorigenesis in mice. Cell. 2024 Feb 15;187(4):882–896.e17.
Li Y, Tong T, Li P, Peng Y, Zhang M, Liu J, et al. Screening of Potential Probiotic Lactobacillaceae and Their Improvement of Type 2 Diabetes Mellitus by Promoting PI3K/AKT Signaling Pathway in db/db Mice. Pol J Microbiol. 2023 Sep 20;72(3):285–297.
Li Z, Lu G, Li Z, Wu B, Luo E, Qiu X, et al. Altered Actinobacteria and Firmicutes Phylum Associated Epitopes in Patients With Parkinson's Disease. Front Immunol. 2021 Jul 2;12:632482.
Kato A, Schleimer RP, Bleier BS. Mechanisms and pathogenesis of chronic rhinosinusitis. J Allergy Clin Immunol. 2022 May;149(5):1491–1503.
Kim SJ, Park JH, Lee SA, Lee JG, Shin JM, Lee HM. All-trans retinoic acid regulates TGF-β1-induced extracellular matrix production via p38, JNK, and NF-κB-signaling pathways in nasal polyp-derived fibroblasts. Int Forum Allergy Rhinol. 2020 May;10(5):636–645.
Huang ZQ, Zhou XM, Yuan T, Liu J, Ong HH, Sun LY, et al. Epithelial Tight Junction Anomalies in Nasal Inverted Papilloma. Laryngoscope. 2024 Feb;134(2):552–561.
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 Nov;256:109791.
Liao S, Huang Y, Zhang J, Xiong Q, Chi M, Yang L, et al. Vitamin D promotes epithelial tissue repair and host defense responses against influenza H1N1 virus and Staphylococcus aureus infections. Respir Res. 2023 Jul 5;24(1):175.
Glaviano A, Foo ASC, Lam HY, Yap KCH, Jacot W, Jones RH, et al. PI3K/AKT/mTOR signaling transduction pathway and targeted therapies in cancer. Mol Cancer. 2023 Aug 18;22(1):138.

No competing interests reported.

Download PDF

Editor assigned by journal
23 Aug, 2024
Submission checks completed at journal
23 Aug, 2024
First submitted to journal
23 Aug, 2024

You are reading this latest preprint version

Identification of host-microbiome interactions in nasal diseases using multiomics integration

Status:

Version 1

Abstract

Background

Objective

Methods

Results

Conclusion

Figures

Introduction

Methods

Subject recruitment and sampling

5R 16S rRNA gene sequencing and analyses

5R 16S rRNA gene sequencing and analyses of negative controls

Bioinformatics analysis of RNA-seq data

Preprocessed microbiome and host transcriptome data

Sparse CCA analysis

Inference of microbial-host interaction networks

Fluorescence in situ hybridization (FISH)

Western blotting

Results

The nasal microbiomes of CRSwNP and NIP patients are distinct

Host transcriptomes of CRSwNP and NIP patients

The disease-specific nasal microbiome is associated with different host pathways

Discussion

Declarations

Consent for publication
Not applicable

Competing interests

Funding

Author Contribution

Acknowledgement

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1

Identification of host-microbiome interactions in nasal diseases using multiomics integration

Status:

Version 1

Abstract

Background

Objective

Methods

Results

Conclusion

Figures

Introduction

Methods

Subject recruitment and sampling

5R 16S rRNA gene sequencing and analyses

5R 16S rRNA gene sequencing and analyses of negative controls

Bioinformatics analysis of RNA-seq data

Preprocessed microbiome and host transcriptome data

Sparse CCA analysis

Inference of microbial-host interaction networks

Fluorescence in situ hybridization (FISH)

Western blotting

Results

The nasal microbiomes of CRSwNP and NIP patients are distinct

Host transcriptomes of CRSwNP and NIP patients

The disease-specific nasal microbiome is associated with different host pathways

Discussion

Declarations

Consent for publication Not applicable

Competing interests

Funding

Author Contribution

Acknowledgement

Data Availability

References

Additional Declarations

Supplementary Files

Status:

Version 1

Consent for publication
Not applicable