Comparison of the nasopharynx microbiome between influenza and non‐influenza cases of severe acute respiratory infections: A pilot study

Abstract Aims Influenza A virus (IAV) can cause severe acute respiratory infection (SARI), and disease outcome may be associated with changes in the microbiome of the nasopharynx. This is a pilot study to characterize the microbiome of the nasopharynx in patients hospitalized with SARI, infected and not infected by IAV. Methods and Results Using target sequencing of the 16S rRNA gene, we assessed the bacterial community of nasopharyngeal aspirate samples and compared the microbiome of patients infected with IAV with the microbiome of patients who were negative for IAV. We observed differences in the relative abundance of Proteobacteria and Firmicutes between SARI patients, with Streptococcus being enriched and Pseudomonas underrepresented in IAV patients compared with patients who were not infected with IAV. Conclusion Pseudomonas taxon seems to be in high frequency on the nasopharynx of SARI patients with non‐IAV infection and might present a negative association with Streptococcus taxon. Microbial profile appears to be different between SARI patients infected or not infected with IAV.


| INTRODUCTION
Influenza A virus (IAV) infection is among the most common and major causes of human respiratory infection, presenting high morbidity and mortality worldwide, with hundreds of thousands of hospitalizations and deaths every year. 1 Hospitalized patients with influenza disease exhibit a variety of nonspecific influenza-like symptoms that may also be observed in patients with other respiratory infections. Hospitalization fatality risk is the probability of death associated with H1N1pdm09 cases in a cohort of individuals that required hospitalization for medical reasons. 2 While the influenza-like symptoms of flu patients are commonly considered as a measure of disease severity, and determine whether the patients suffer from a severe acute respiratory infection (SARI), hospitalization fatality risk during influenza virus infection has been underestimated. 2 Thus, the addition of other measures that could impact severity-such as the microbiome-should be explored.
The microbiome could be defined as the collective genome of the microorganisms that reside in an environment niche. 3,4 Studies of the human microbiome have shown a remarkable diversity of microbes that occupy different habitats of the human body to establish a microbial community. 5,6 These microbial communities seem to be structurally stable over time, and this stability of the microbiome composition has been associated with specific behaviors of individuals, such as observed on healthy smokers, 7 as well as to health condition of individuals, such as observed on patients with cystic fibrosis 8 or infectious disease. 9 The respiratory tract has been widely studied to understand the dynamics of respiratory infections. 10 While lung samples are not easily accessible, nasal and oral samples have been used for investigating and identifying microorganisms responsible for lung infection. 10,11 Recently, comparative studies have shown that the bronchoalveolar microbiota may be better represented by a composition of oral and nasal microbiomes. 11 However, IAV H1N1 subtype is a respiratory virus, and its transmission typically comprises airway introduction and success infection of the upper respiratory tract (URT). 12 The microbiome of the URT is susceptible to disruption by pathogens. Influenza A virus infection, for example, has been shown to modify the community structure of the microbiome 13 and to lead to the outgrowth of pathogenic bacteria. 14 Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria, and Fusobacteria are common phyla found in variable proportions in the URT of healthy individuals. 15 Even pathogenic bacteria can be present at low abundance in established communities. 5 A specific ecological perturbation can, however, change the bacterial community structure, leading to local or systemic infection by both bacterial and viral pathogens, 16  informed that their samples and the health-related data collected would be used for disease diagnosis, clinical treatment, and epidemiological surveillance and that the data could be further used for scientific research. Patients were given the opportunity to refuse, and only data from patients who agreed with these terms were included in the study.
All data were analyzed and reported anonymously and kept confidential. The authors did not have access to identifiers of research subjects other than clinical data, sex, age, and pregnancy and vaccination status. Samples were considered eligible when medical records indicated no smoking behavior, 7 no previous vaccine for IAV H1N1pdm09/ H3N2, 19 no comorbidities such as chronic pneumopathy or chronic heart disease, nonchronic viral diseases such as hepatitis C or HIV infection, and negative for other respiratory viruses.  All 16S rRNA gene reads produced by high-throughput sequencing were subjected to quality control to retain sequences with a minimum length of 100 bp and were trimmed to remove low-quality bases (minimum Phred score of 30) using PRINSEQ. 23 Also, duplicated sequences were identified and sorted by decreasing read abundance and then filtered to exclude singletons, using USEARCH v7.0.1090. 24 Clusters were assembled using a minimum identity of 99%, and chimeras were removed using the RDP reference database. 25 Taxonomic assignment was obtained using QIIME v1.8.0, 26 and operational taxonomic units (OTUs) were selected on the basis of 97% sequence similarity. Taxonomic data were generated through the classification algorithm using the 97% OTUs version of GreenGenes 13.8. 27 The default parameters of QIIME v1.8.0 were used for the alignment of OTUs (pyNAST) and to generate phylogenies (FastTree). Rarefactions of the OTU table were performed on 10 steps of 500 sequences of subsampling for a maximum depth of 5000 sequences. Alpha diversity metrics were calculated using QIIME v1.8.0. Multiple rarefactions were performed for Chao1 (species richness), Shannon (the entropic information of the abundances of observed OTUs), Simpson_e (evenness), and Equitability. Beta diversity analysis was calculated using unweighted UniFrac. The principal coordinate analysis (PCoA) was generated to observe differences between groups, and the results were visualized using EMPeror software. 28

| Statistical analysis
Statistical analyses were done using SPSS 20.0 (IBM, USA). Data were presented as relative frequency or median and interquartile ranges.
The Mann-Whitney U test was used to compare the diversity between groups. Values were considered statistically significant when P < .05 (2-tailed test).

| The genus Pseudomonas is associated with non-IAV hospitalized patients
The taxonomic classification of the sequences for these 12 samples revealed the nasopharynx to be colonized by 9 bacterial phyla, albeit not all simultaneously. A comparison between IAV patients and non-IAV SARI patients showed significant differences in the frequencies of Proteobacteria and Firmicutes.
At 97% similarity, the sequences matched 110 different OTUs, from which 52 (47.2%) had frequencies above 1% of the total reads.
Ten of the 12 samples had at least one-third of the total reads overrepresented by 1 genus. In general, the most abundant bacterial genera found across samples were Prevotella (at an average relative abundance of 15.8%), Pseudomonas (11.7%), and Streptococcus (9.5%). Significant differences between IAV and non-IAV groups were seen for 15 genera (Figure 1). At all taxonomic levels, the sequences that could not be classified to known taxa ranged from 1.4% to 16.9% of the total reads, and no significant differences were observed.
Interestingly, the bacterial genus Pseudomonas appeared to be absent-or present at a very low relative abundance (0.01% and 0.06% in 2/6)-in samples from IAV patients, while it was present at a high relative abundance in 5 of the 6 samples from non-IAV patients  In this study, both IAV patients and non-IAV patients exhibited a great bacterial diversity in the nasopharynx, and the average number  An association between IAV and Pseudomonas was previously reported, suggesting that IAV infection may facilitate the establishment of this pathogenic bacterium in the lower respiratory tract. 33 In our study, however, we did not find evidence for such association. In fact, the genus Pseudomonas was identified in only 2 of 6 IAV samples (IP5 and IP6) and at a very low relative abundance (less than 1%), whereas it was found at a high relative abundance in all non-IAV samples. Pseudomonas is a bacterial genus related to several human infections, and it has been considered an opportunistic pathogen present in the respiratory tract of humans. 34 Some species of this genus are able to produce a biofilm and express flagellum protein as well as several other adhesins, such as pili. These characteristics are important for colonization and adhesion to mucins, glycoproteins found in airway mucus. 35 Pseudomonas has been shown to induce host expression of MUC2 Our results suggest a trend that would need to be confirmed with a larger number of specimens to determine whether a specific microbiota in the URT that is associated with severity of disease indeed exists, such that could be predictive of poor outcomes in patients infected with influenza. On the basis of our findings, we suggest that the presence of Streptococcus is not necessarily indicative of poor outcome in IAV patients, but that shifts in its relative abundance, or even concomitance of its presence with the presence or absence of other specific species, may be. Our findings also suggest that in patients with SARI, Pseudomonas and IAV are not always found in association.