SARS-CoV-2 Variant-Dependent Alterations in Nasopharyngeal Microbiota and Host Inflammatory Response

Abstract

The SARS-CoV-2 pandemic saw multiple outbreaks occur over short periods. This was linked to the viruss high infectivity and rapid mutation rate, which hindered the development of effective treatments and a comprehensive understanding of COVID-19 pathophysiology. The nasopharyngeal tract, the main entry site of SARS-CoV-2, interacts with angiotensin-converting enzyme 2 (ACE2) receptors via the viral spike glycoprotein, triggering pro-inflammatory responses that affect both tissue integrity and the resident microbiota. Viral infection induces dysbiosis by modulating microbiome diversity. Although distinct variants produce independent symptoms and viral loads, their impact on the nasopharyngeal microbiota and host inflammatory profile remains poorly understood. To address this, we analyzed nasopharyngeal samples from Chilean individuals collected during different phases of the pandemic. The microbiota was characterized by 16S rRNA gene sequencing (V3-V4 region), and cytokine expression was quantified by RT-qPCR. Finally, applied rigorous data processing and machine-learning models Random Forest and K-Nearest Neighbors to identify associations between SARS-CoV-2 variants, inflammatory markers, and opportunistic bacterial genera. Our results reveal that SARS-CoV-2 infection promotes distinct immunomicrobial signatures marked by TNF--driven inflammation and the expansion of opportunistic taxa. These variant-dependent alterations indicate that host inflammatory responses and microbial dysbiosis are closely intertwined in the nasopharyngeal environment. This study provides a comprehensive framework integrating microbial and host factors to better understand the mechanisms underlying COVID-19 pathogenesis and highlights the potential of combining microbiome and cytokine profiling with machine-learning approaches to differentiate infection outcomes across viral variants. ImportanceUnderstanding how SARS-CoV-2 infection alters the upper respiratory microbiota and host immune responses is essential to uncover the mechanisms underlying COVID-19 severity. In this study, we integrate bacterial 16S rRNA sequencing, inflammatory markers, and machine learning to identify key descriptors of infection such as TNF-, Acinetobacter, Prevotella, and Staphylococcus. These markers define a conserved immunomicrobial signature across viral variants, linking inflammatory activation with microbial dysbiosis in the nasopharyngeal environment. Our findings highlight the value of the correct and robust application of machine learning models and nested cross-validation strategies in high-dimensional (wide) datasets to improve variant discrimination, reveal host-microbe interactions, and support the development of therapeutic strategies for respiratory viral infections.