Gut microbial allies and adversaries in SARS-CoV-2 infection Authors: Veronica Tolosa-Enguis, María Leonor Fernández-Murga, Alfonso Benitez-Páez, Carlos Francés-Cuesta, Javier Garde-Noguera, Antonio Llombart-Cusscac, Gara Verónica Molina-Mendoza, Ana Isabel Aviles-Alia, Ron Geller, Laura Mendoza-García, María Montoya, Eva María Medina-Rodríguez, Yolanda Sanz Published: 2026 ABSTRACT Highlights•COVID-19 continues to circulate at high levels and is still an important public health concern.•The gut microbiota plays an important role but its therapeutic potential remains unexplored.•This cross-sectional study shows the link between COVID-19 and the gut microbiota.•We propose gut bacterial candidates which might help manage the disease.•This information would enhance preparedness for future viral threats. Introduction Coronavirus disease 2019 (COVID-19) is a highly contagious respiratory disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Although most cases of COVID-19 are asymptomatic or mild, the disease can progress to severe clinical illness, respiratory failure, and even death, particularly in immunocompromised and older people (Rossini et al., 2024). The first case was detected in Wuhan in December 2019, and the World Health Organization (WHO) declared it a pandemic less than three months later (Zhu et al., 2020, Yamamoto et al., 2021, Stawicki et al., 2020). Since the start of the pandemic, over 780 million reported COVID-19 cases and over 7.1 million COVID-19 deaths worldwide (World Health Organization 2026). Like all viruses, SARS-CoV-2 evolves, and some mutations can significantly affect its transmissibility and virulence, as well as vaccine effectiveness. Due to the virus’s evolving nature, the ongoing regional COVID-19 cases, and the pandemic potential of coronaviruses in general, WHO continues to monitor COVID-19 alongside other serious coronavirus-related diseases such as severe acute respiratory syndrome (SARS) and Middle East respiratory syndrome (MERS) through global surveillance initiatives like CoViNet (World Health Organization 2024). Several studies have reported digestive system changes in COVID-19 patients, such as diarrhoea, liver function changes, and even multi-organ failure (Huang et al., 2020, Chen et al., 2020, Cheung et al., 2020). The intestinal tract contributes to SARS-CoV-2 pathogenicity because the virus can replicate within intestinal enterocytes. These cells, like lung cells, express the angiotensin-converting enzyme 2 (ACE2) receptor, the primary viral entry point that recognizes the SARS-CoV-2 spike (S) protein. Specifically, this interaction is mediated by the S protein’s receptor binding domain (RBD) (Rossini et al., 2024, Lamers et al., 2020, Shang et al., 2020, Gargantilla et al., 2023). The ACE2 receptor also plays an important regulatory role in dietary amino acid homeostasis, innate immunity, antimicrobial peptide expression and, most importantly, intestinal ecology (Hashimoto et al., 2012, Ferreira-Duarte et al., 2021, Guo et al., 2021, Schirmer et al., 2016, Diab et al., 2025). As the site of the body’s largest population of immunocompetent cells, the gut – together with its microbiota – plays a key role in regulating local and systemic immune responses (Rossini et al., 2024, Troisi et al., 2021, Han et al., 2020, Wittig and Zeitz, 2003). Several studies have demonstrated a close relationship between gut immunity and COVID-19. For example, studies using intestinal organoids have shown that enterocytes are a pro-inflammatory reservoir for the virus. In the gut, SARS-CoV-2 disrupts the innate immune response and interferes with the autocrine action of interferons (IFNs) in infected enterocytes, thereby enhancing viral replication (Lamers et al., 2020). In addition, viral SARS-CoV-2 ribonucleic acid (RNA) has been detected in faecal samples of infected individuals with altered gut microbiota (Lamers et al., 2020, Gargantilla et al., 2023, Zuo et al., 2020). On one hand, the gut microbiota composition may influence susceptibility to infection (Ng et al., 2023), and on the other hand, SARS-CoV-2 infection alters the gut microbiota (Gaibani et al., 2021), potentially impacting disease severity and long-term symptoms (Troisi et al., 2021, Ng et al., 2023). Several studies have shown that people with COVID-19 exhibit reduced gut microbiota richness and diversity, alongside an increased abundance of potential pathobionts (Koupaei et al., 2024). It may take up to six months for the microbiota to recover to its pre-infection state (Zuo et al., 2020, Chen et al., 2022). In particular, butyrate-producing commensal species from the Ruminococcaceae and Lachnospiraceae families are less abundant in infected individuals (Ng et al., 2023, Gu et al., 2020, Nogal et al., 2021). Bacteria with immunomodulatory properties, such as Faecalibacterium prausnitzii, Eubacterium rectale, and several species of Bifidobacterium, are also depleted (Rossini et al., 2024, Yeoh et al., 2021). In addition, comorbidities frequently associated with severe COVID-19 (obesity, coronary artery disease, hypertension, type-2 diabetes) are themselves associated with gut microbiota alterations, especially in Bacteroidota and Bacillota phyla (formerly Bacteroidetes and Firmicutes phyla, respectively) (Turnbaugh et al., 2006, Emoto et al., 2016, zhi et al., 2024, Yang et al., 2015, Yan et al., 2022). COVID-19 patients also show changes in microbiome functionality, including increases in de novo nucleotide biosynthesis, amino acid biosynthesis, and glycolysis (Wang et al., 2023). Furthermore, enterocyte infection can trigger inflammation, and this response, combined with virus-induced gut dysbiosis, can modify gut barrier permeability (Han et al., 2020, Xu et al., 2021) and alter the immune response, increasing serum cytokine levels (Xu et al., 2021). Exacerbated immunological response to SARS-CoV-2 infection, known as a cytokine storm, is a leading cause of respiratory failure and death in severe COVID-19. A protective immunoregulatory gut microbiota could theoretically prevent or modulate this response (Rossini et al., 2024). Additionally, some previous research suggests a bidirectional relationship between the gut and respiratory tract microbiota through the “gut-lung axis”, enabling the exchange of immune cells and microorganisms between these organs (Xu et al., 2021, Chen et al., 2021, Sweeney and Morton, 2013). Other viral respiratory infections, such as influenza, may be associated with gut microbiota alterations that predispose patients to secondary lung infections (Xu et al., 2021, Cenit et al., 2017, Rob et al., 2025, Capuco et al., 2020). For all these reasons, elucidating the host-microbial perturbations underlying SARS-CoV-2 infection could be crucial for improving the efficacy of future interventions, including vaccines, antivirals, and probiotic adjuvants. We aimed to advance our understanding of the relationships between gut microbiota, altered biochemical parameters, and disease severity in people hospitalized with COVID-19. We also investigated the immunomodulatory and anti-infectivity roles of differentially abundant bacteria in this patient population. Materials and methods Study design and sampling We conducted a cross-sectional study at Hospital Arnau de Vilanova-Lliria (Valencia, Spain) between October and December 2020, during the outbreak of the Wuhan variant. Our sample comprised 63 hospitalized individuals with SARS-CoV-2 infection and 54 non-infected controls. Briefly, the inclusion criteria were: 1) adults (age ≥ 18 years) admitted to Hospital Arnau de Vilanova-Lliria between October and December 2020, 2) positive nasal and/or pharyngeal swab for SARS-CoV-2 RNA confirmed by RT-qPCR assay (AllplexTM Next Generation of Real-Time PCR Assay, Seegene Inc., Seoul, South Korea), 3) complete clinical and laboratory data available at the time of admission, and 4) ability to give written consent. We excluded people who were taking part in other clinical trials on COVID-19. The control cohort consisted of age-matched healthy volunteers, recruited from October to December 2020, with no history of infection (no positive serology (immunoglobulin (Ig)M/IgG) for SARS-CoV-2), no acute infection at the time of sampling, and no use of antibiotics in the month prior to sample collection. It was not possible to recruit antibiotic-free patients, because the hospital protocol stipulated antibiotic administration to people with confirmed or suspected SARS-CoV-2 infection as a precaution against bacterial co-infection or superinfection, particularly in severe cases. The COVID-19 patients were classified into four severity groups (mild, moderate, severe, and critical) based on established clinical criteria (Enaud et al., 2020). Demographic, clinical, biochemical, and immunological data were obtained from medical records and additional analyses. Biochemical and hematological parameters were assessed as part of the clinical characterization of the SARS-CoV-2 study. Results were summarized descriptively and interpreted using the reference intervals established or verified by the laboratory performing each assay. Blood and stool samples were collected during hospitalization, and SARS-CoV-2-specific IgM and IgG levels were assessed by enzyme-linked immunosorbent assay (ELISA; LIASON® SARS-CoV-2 S1/S2 IgG test and LIASON® SARS-CoV-2 IgM test, DiaSorin, Saluggia, Italy) two days and two months after infection. Hospitalization duration was calculated as the number of days from hospital admission to discharge, or to death in the case of the deceased patient. Asymptomatic hospitalized patients were included in this calculation because they were admitted for precautionary reasons due to advanced age and/or relevant comorbidities. The study was approved by the Ethics Committee of Arnau de Vilanova-Lliria Hospital (HAV-BAR-2020-03) and of the CSIC Ethics Committee (Ref. 137/2020) and was conducted in accordance with Spanish and European legislation, with informed consent obtained from all participants and anonymity guaranteed. The Supplementary material provides more detailed information on participant recruitment and characteristics. Microbiota analysis Stool samples were processed for DNA extraction with the QIAamp® Fast DNA Stool Mini Kit (Qiagen, Germany). Processing included initial viral inactivation via incubation at 70 °C for 10 minutes. Bacterial DNA was amplified by targeting the full-length ribosomal RNA (rRNA) operons (∼4.5 kb). Briefly, dual-barcoded polymerase chain reaction (PCR) products were amplified in duplicate by 30 PCR cycles at 95 °C for 30 seconds, 50 °C for 20 seconds, and 72 °C for 240 seconds. The PCR products were sequenced on the MinION Mk1C platform (Oxford Nanopore Technologies). The FASTQ raw data for amplicon pools generated in this study can be publicly accessed through the PRJEB102567 bioproject at the European Nucleotide Archive. Basecalling, quality control, and demultiplexing were performed using Guppy, FastQC, Cutadapt, and custom scripts. The reads were mapped against the MirROR database, enabling species-level taxonomic profiling. Alpha and beta diversity metrics were then calculated using QIIME and R libraries (CoDaSeq, vegan and Zcompositions), with compositional data transformations and non-metric multidimensional scaling (NMDS) analysis based on Aitchison distances. Several potential confounders were considered, including age, sex, body mass index (BMI), pre-existing conditions, batch, and medication use. The Supplementary material provides full descriptions of all experimental protocols and bioinformatics pipelines. Functional evaluation of gut bacteria in infection models To explore the immunomodulatory effects of gut bacterial species identified as potentially relevant to infection in our observational study, we used the following selected strains: Bacteroides uniformis IATA25 (own collection, B. uniformis IATA25), Agathobacter faecis (formerly Roseburia faecis) DSM16840, Romboutsia lituseburensis DSM797, Barnesiella intestinihominis DSM21032, Fusicatenibacter saccharivorans DSM26062, and Enterococcus faecalis CECT481. All bacteria were killed with ultraviolet radiation and then exposed to human peripheral blood mononuclear cell cultures (PBMCs; STEMCELL Technologies) that were stimulated with SARS-CoV-2 peptide pools (PepMixTM SARS-CoV-2 –S-MUT; JPT Peptide Technologies–) or with pseudotyped viruses (vesicular stomatitis virus (VSV) pseudotyped with S protein of the SARS-CoV-2 XBB.1.5 variant (VSV-XBB.1.5) and VSV pseudotyped with its native glycoprotein (VSV-G) as control). Cytokine levels (tumour necrosis factor (TNF), interleukin (IL)-12P70, IL-5, IL-6, IL-4, interferon-gamma (IFN-γ)) were quantified using Luminex ProcartaPlex™ Human Essential Th1/Th2 Cytokine Panel Kit (Thermo Fisher Scientific) and ELISA kits (IL-4 Human Uncoated ELISA Kit, Thermo Fisher Scientific). To evaluate the balance between anti- and pro-inflammatory responses in PBMC stimulation assays, a relative anti-inflammatory index was calculated for each condition as the ratio of IL-4 to the corresponding pro-inflammatory cytokine (TNFα, INFγ, or IL-6) in the presence of viral antigens plus bacterium, normalized to the baseline ratio obtained for the bacterium alone: Anti−inflammatoryratio=(IL−4Pro−inflammatorycytokine)Antigens+Bacterium(IL−4Pro−inflammatorycytokine)Bacteriumalone. Values freater than 1 denote a net shift toward an anti-inflammatory cytokine profile induced by the presence of viral antigens and the evaluated bacterial strain. Additionally, VeroE6-TMPRSS2 (JCRB1819, Japanese Collection of Research Bioresources) cells were employed in in vitro infection models to evaluate the impact of bacterial exposure on susceptibility to SARS-CoV-2 spike-pseudotyped viral entry as measured by green fluorescent protein (GFP) fluorescence. The Supplementary material provides full details. Production of short-chain fatty acids (SCFAs) by selected gut bacterial strains To evaluate the ability of gut bacterial species inversely associated with COVID-19 infection to produce immunomodulatory metabolites such as short-chain fatty acids (SCFAs), we cultured the strain B. uniformis IATA25 in Schaedler medium (02-599-500, Scharlau) for 20 h at 37°C under anaerobic conditions, and we cultured E. faecalis CECT481 in brain-heart infusion (BHI) broth (CM0497B, OXOID) for 8 h under aerobic conditions with agitation. Following incubation, bacterial viability was assessed by flow cytometry. The cultures were then centrifuged at 2000 × g for 15 min at 4°C, and the resulting supernatants were concentrated 20-fold by lyophilization. Short-chain fatty acids (SCFAs) were quantified by gas chromatography-mass spectrometry (GC-MS) according to the protocol described by Eberhart et al. (Eberhart et al., 2021). Briefly, calibration curves were prepared for acetic (71251, Sigma-Aldrich), propionic (94425, Sigma-Aldrich), and butyric acids (19215, Sigma-Aldrich). A volume of 200 µL per sample was mixed with 1 mL of diethyl ether (ET00831000, Scharlau) and anhydrous Na₂SO₄ (238597, Sigma-Aldrich) to remove residual water, incubated for 10 min, and centrifuged at 2000 × g for 5 min. The organic phase was injected into an Agilent G3903-63008 GC–MS system equipped with a Gerstel MPS autosampler and an Agilent DB-FATWAX UI column (30 m × 250 µm × 0.25 µm). The SCFAs were quantified using MassHunter software. Statistical analysis Statistical analyses were conducted using the most appropriate tests based on data normality according to the Shapiro-Wilk test. For the microbiota data, Wilcoxon Rank Sum test, PERMANOVA, generalized linear mixed models (GLMs), and Kendall's rank correlations were applied to identify associations between bacterial abundance and clinical variables. For in vitro functional assays, parametric data were analysed using Student’s t-test or one-way analysis of variance (ANOVA) followed by Tukey’s post-hoc for multiple comparisons. False discovery rate (FDR) corrections (Benjamini-Hochberg) were applied to account for multiple comparisons, and results with p-values or q-value below 0.05 were considered statistically significant. The Supplementary material provides full details. Results Participant characteristics Table 1 presents the COVID-19 patients’ demographic and clinical features, and Table 2 presents the controls’ demographic and physiological features. The COVID-19 group included 54% males and 46% females and had a median age of 59 years (range 15 to 92 years), compared with 43.5% males and 56.5% females and a median age of 54 years (range 15 to 89 years) in the control group. Comorbidities were more prevalent and diverse in the COVID-19 group. Most COVID-19 patients (63.5%) had mild disease, five (7.9%) required intensive care unit admission, and one died with SARS-CoV-2 infection. Among survivors, the median duration of symptoms from diagnosis was six days (range 4 to 10.5 days).Table 1Demographic and clinical features of COVID-19 patients.DemographicsN (%)aAge (median), [range]59 [15-92]Male34 (54%)Female29 (46%)SymptomsMild40 (63.5%)Moderate10 (15.9%)Severe8 (12.7%)Critical3 (4.7%)Asymptomatic2 (3.2%)OutcomesHospitalization63 (100%)Intensive Care Unit5 (7.9%)Death1 (1.6%)Symptoms at admissionCough45 (71%)Fever41 (65%)Dyspnoea31 (49%)Odynophagia4 (6%)Vomits5 (8%)Diarrhoea18 (29%)Asthenia39 (62%)Anosmia17 (27%)Headache16 (25%)Duration of the symptomsDays (median), [range]6 [4-10.5]Clinical complications due to COVID-19Pneumonia55 (87%)Sepsis5 (7.9%)Heart failure4 (6%)Arrhythmia4 (6%)Respiratory distress syndrome20 (31.7%)Treatment for COVID-19Antibioticsb60 (95%)Antifungals4 (6%)Antivirals (Remdesivir)10 (15.9%)Steroids (Dexametasone)30 (47.6%)Anti-IL6 (Tocilizumab)9 (14%)Anticoagulant (LMWH)c59 (93.7%)Ventilation4 (6%)Othersd24 (38%)ComorbiditiesHypertension24 (38%)Dyslipidaemia20 (31.7%)Diabetes8 (12.7%)Cardiovascular disease12 (19%)Renal disease11 (17.5%)Respiratory problems/ Lung disease12 (19%)Others41 (65%)Qualitative BMIUnderweight3 (4,76%)Normal12 (19.05%)Overweight27 (42.86%)Obese21 (33.33%)Smoking HabitsFormer smoker14 (22.2%)Smoker6 (9.5%)Non-smoker42 (66.7%)NAe1 (1.6%)Frequent alcohol consumptionYes7 (11.1%)No55 (87.3%)NAe1 (1.6%)SeroprevalencefIgM-/IgG+12 (19%)IgM+/IgG+32 (50.8%)IgM-/IgG-1 (1.6%)UDg18 (28.6%)aData are expressed as the number of observations, with the percentage of total population given in brackets.bMainly cephalosporin (n=42, 66.7%).cLMWH: Low Molecular Weight HeparindTuberculosis, dementia, anxiety, osteoarthritis, stroke, psoriasis, etc.eNA: data not availablefThe determination was performed between 2 and 30 days after the first COVID-19 positive PCRgUD: undeterminedHospitalization duration was calculated as the number of days from admission to discharge or death. Asymptomatic patients (admitted for precautionary reasons due to advanced age/comorbidities) and the deceased patient were included in this calculation.Table 2Demographic and physiological features of the controls.DemographicsN (%)aAge (median), [range]54 [15-89]Male25 (36.5%)Female29 (56.5%)MedicationAntibiotics5 (10.2%)Othersb24 (49%)ComorbiditiesHypertension4 (8%)Dyslipidaemia11 (22.4%)Diabetes2 (4.1%)Cardiovascular disease4 (8.2%)Renal disease4 (8.2%)Respiratory problems/Lung disease15 (30.6%)Others11 (22.4%)Qualitative BMINormal22 (44%)Overweight20 (40%)Obese8 (16%)Smoking HabitsFormer smoker1 (2%)Smoker4 (8%)Non-smoker27 (54%)NAc18 (36%)Frequent alcohol consumptionYes24 (44.44%)No26 (48.15%)NAc4 (7.41%)aData are expressed as the number of observations, with the percentage of total population given in brackets.bAnti-inflammatories, antihistamines, antipsychotics, antidepressants, benzodiazepines...cNA: data not available Table 1Demographic and clinical features of COVID-19 patients.DemographicsN (%)aAge (median), [range]59 [15-92]Male34 (54%)Female29 (46%)SymptomsMild40 (63.5%)Moderate10 (15.9%)Severe8 (12.7%)Critical3 (4.7%)Asymptomatic2 (3.2%)OutcomesHospitalization63 (100%)Intensive Care Unit5 (7.9%)Death1 (1.6%)Symptoms at admissionCough45 (71%)Fever41 (65%)Dyspnoea31 (49%)Odynophagia4 (6%)Vomits5 (8%)Diarrhoea18 (29%)Asthenia39 (62%)Anosmia17 (27%)Headache16 (25%)Duration of the symptomsDays (median), [range]6 [4-10.5]Clinical complications due to COVID-19Pneumonia55 (87%)Sepsis5 (7.9%)Heart failure4 (6%)Arrhythmia4 (6%)Respiratory distress syndrome20 (31.7%)Treatment for COVID-19Antibioticsb60 (95%)Antifungals4 (6%)Antivirals (Remdesivir)10 (15.9%)Steroids (Dexametasone)30 (47.6%)Anti-IL6 (Tocilizumab)9 (14%)Anticoagulant (LMWH)c59 (93.7%)Ventilation4 (6%)Othersd24 (38%)ComorbiditiesHypertension24 (38%)Dyslipidaemia20 (31.7%)Diabetes8 (12.7%)Cardiovascular disease12 (19%)Renal disease11 (17.5%)Respiratory problems/ Lung disease12 (19%)Others41 (65%)Qualitative BMIUnderweight3 (4,76%)Normal12 (19.05%)Overweight27 (42.86%)Obese21 (33.33%)Smoking HabitsFormer smoker14 (22.2%)Smoker6 (9.5%)Non-smoker42 (66.7%)NAe1 (1.6%)Frequent alcohol consumptionYes7 (11.1%)No55 (87.3%)NAe1 (1.6%)SeroprevalencefIgM-/IgG+12 (19%)IgM+/IgG+32 (50.8%)IgM-/IgG-1 (1.6%)UDg18 (28.6%)aData are expressed as the number of observations, with the percentage of total population given in brackets.bMainly cephalosporin (n=42, 66.7%).cLMWH: Low Molecular Weight HeparindTuberculosis, dementia, anxiety, osteoarthritis, stroke, psoriasis, etc.eNA: data not availablefThe determination was performed between 2 and 30 days after the first COVID-19 positive PCRgUD: undeterminedHospitalization duration was calculated as the number of days from admission to discharge or death. Asymptomatic patients (admitted for precautionary reasons due to advanced age/comorbidities) and the deceased patient were included in this calculation. Demographic and clinical features of COVID-19 patients. Data are expressed as the number of observations, with the percentage of total population given in brackets. Mainly cephalosporin (n=42, 66.7%). LMWH: Low Molecular Weight Heparin Tuberculosis, dementia, anxiety, osteoarthritis, stroke, psoriasis, etc. NA: data not available The determination was performed between 2 and 30 days after the first COVID-19 positive PCR UD: undetermined Hospitalization duration was calculated as the number of days from admission to discharge or death. Asymptomatic patients (admitted for precautionary reasons due to advanced age/comorbidities) and the deceased patient were included in this calculation. Table 2Demographic and physiological features of the controls.DemographicsN (%)aAge (median), [range]54 [15-89]Male25 (36.5%)Female29 (56.5%)MedicationAntibiotics5 (10.2%)Othersb24 (49%)ComorbiditiesHypertension4 (8%)Dyslipidaemia11 (22.4%)Diabetes2 (4.1%)Cardiovascular disease4 (8.2%)Renal disease4 (8.2%)Respiratory problems/Lung disease15 (30.6%)Others11 (22.4%)Qualitative BMINormal22 (44%)Overweight20 (40%)Obese8 (16%)Smoking HabitsFormer smoker1 (2%)Smoker4 (8%)Non-smoker27 (54%)NAc18 (36%)Frequent alcohol consumptionYes24 (44.44%)No26 (48.15%)NAc4 (7.41%)aData are expressed as the number of observations, with the percentage of total population given in brackets.bAnti-inflammatories, antihistamines, antipsychotics, antidepressants, benzodiazepines...cNA: data not available Demographic and physiological features of the controls. Data are expressed as the number of observations, with the percentage of total population given in brackets. Anti-inflammatories, antihistamines, antipsychotics, antidepressants, benzodiazepines... NA: data not available We measured COVID-19 patients’ serum antibody levels (two days to two months after testing positive) and found that 44 patients (69.8%) had seroconverted, 32 (50.8%) were double positive, and 12 (19%) were IgM-negative/IgG-positive, while only one (1.6%) was double negative. The COVID-19 patients showed significant increases in ferritin, alanine aminotransferase (ALT), lactate dehydrogenase (LDH), c-reactive protein (CRP), D-dimer, and IL-6 compared to established reference values, whereas vitamin D levels were decreased (reference range: 30–100 ng/mL) (Table 3). Their CD4+ lymphocyte counts were above established reference values, with a median value of 51.3% (Table 4). We found no association between disease severity and other variables such as sex; age; and presence of diarrhoea, vomiting, asthenia, anosmia, or headache (data not shown).Table 3Biochemical parameters of COVID-19 patients.VariableaReference valuesMean95% CIMedianStandard DeviationStandard Error of the Mean (SEM)Ferritin15-150 ng/mL584.4462.2-706.6485481.161.11Vitamin D30-100 ng/mL24.4621.0-27.92413.31.71D-dimer0-500 ng/mL1251146.5-2355.1503431.17552IL-60-7 pg/mL40.64NA16.5511414.98HaemoglobinFemale: 12-16 g/dLFemale: 12.54Female: 11.9-13.1Female: 12.54Female: 1.46Female: 0.23Male: 13-18 g/dLMale: 13.89Male: 13.4-14.7Male: 14.25Male: 1.38Male: 0.27ALT0-33 mg/dL29.923.8-35.92023.83.02AST0-32 mg/dL36.3726.6-46.124.537.84.88LDH0-250 mU/mL269.8196.9-342.8220277.436.43Glucose74-109 mg/dl102.795.7-109.79627.83.5Triglycerides0-199 mg/dL118.8106.1-131.510950.46.35CreatinineFemale: 0.59-1.04 mg/dLFemale: 0.95Female: 0.6-0.8Female: 0.85Female: 0.26Female: 0.07Male: 0.74-1.34 mg/dLMale: 0.71Male: 0.8-1.1Male: 0.66Male: 0.45Male: 0.04CRP0-5 mg/L58.9245.0-72.848.355.26.94Total IgG700-1600 mg/dL1067.9927.4-1208.4982543.870.20Total IgM40-230 mg/dL11085.8-134.28593.812.11Total IgA40-400 mg/dL253.1222.9-283.3241.5117.0315.11aAST: aspartate transaminase; ALT: alanine aminotransferase; LDH: lactate dehydrogenase; CRP: C-reactive protein.Biochemical parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values.Table 4Blood cell profiles of COVID-19 patients.VariableReference valuesMean95% CIMedianStandard DeviationStandard Error of the Mean (SEM)Leukocytes (total)4-11.5 × 1095.65.0-6.34.92.60.32Neutrophils1.5-7.5 × 1093.83.2-4.53.20.50.32Lymphocytes (total)1-4 × 1091.21.1-1.31.20.50.004Basophils0-0.4 × 1090.00.0-0.00.00.00.32Platelets125-450 × 10921.7193.8-240.1199.591.111.56% T cells (total)65-75%66.363-697011.81.55% B cells10-15%11.39.6-12.9106.10.81% CD4+ T cells35-45%51.331.6-71.14171.59.86% CD8+ T cells30-40%21.819.5-24.2209.01.18% Natural killers10-20%20.117.0-23.21811.61.54Hematological parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values. Table 3Biochemical parameters of COVID-19 patients.VariableaReference valuesMean95% CIMedianStandard DeviationStandard Error of the Mean (SEM)Ferritin15-150 ng/mL584.4462.2-706.6485481.161.11Vitamin D30-100 ng/mL24.4621.0-27.92413.31.71D-dimer0-500 ng/mL1251146.5-2355.1503431.17552IL-60-7 pg/mL40.64NA16.5511414.98HaemoglobinFemale: 12-16 g/dLFemale: 12.54Female: 11.9-13.1Female: 12.54Female: 1.46Female: 0.23Male: 13-18 g/dLMale: 13.89Male: 13.4-14.7Male: 14.25Male: 1.38Male: 0.27ALT0-33 mg/dL29.923.8-35.92023.83.02AST0-32 mg/dL36.3726.6-46.124.537.84.88LDH0-250 mU/mL269.8196.9-342.8220277.436.43Glucose74-109 mg/dl102.795.7-109.79627.83.5Triglycerides0-199 mg/dL118.8106.1-131.510950.46.35CreatinineFemale: 0.59-1.04 mg/dLFemale: 0.95Female: 0.6-0.8Female: 0.85Female: 0.26Female: 0.07Male: 0.74-1.34 mg/dLMale: 0.71Male: 0.8-1.1Male: 0.66Male: 0.45Male: 0.04CRP0-5 mg/L58.9245.0-72.848.355.26.94Total IgG700-1600 mg/dL1067.9927.4-1208.4982543.870.20Total IgM40-230 mg/dL11085.8-134.28593.812.11Total IgA40-400 mg/dL253.1222.9-283.3241.5117.0315.11aAST: aspartate transaminase; ALT: alanine aminotransferase; LDH: lactate dehydrogenase; CRP: C-reactive protein.Biochemical parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values. Biochemical parameters of COVID-19 patients. AST: aspartate transaminase; ALT: alanine aminotransferase; LDH: lactate dehydrogenase; CRP: C-reactive protein. Biochemical parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values. Table 4Blood cell profiles of COVID-19 patients.VariableReference valuesMean95% CIMedianStandard DeviationStandard Error of the Mean (SEM)Leukocytes (total)4-11.5 × 1095.65.0-6.34.92.60.32Neutrophils1.5-7.5 × 1093.83.2-4.53.20.50.32Lymphocytes (total)1-4 × 1091.21.1-1.31.20.50.004Basophils0-0.4 × 1090.00.0-0.00.00.00.32Platelets125-450 × 10921.7193.8-240.1199.591.111.56% T cells (total)65-75%66.363-697011.81.55% B cells10-15%11.39.6-12.9106.10.81% CD4+ T cells35-45%51.331.6-71.14171.59.86% CD8+ T cells30-40%21.819.5-24.2209.01.18% Natural killers10-20%20.117.0-23.21811.61.54Hematological parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values. Blood cell profiles of COVID-19 patients. Hematological parameters are presented as descriptive clinical characteristics of the SARS-CoV-2 cohort and interpreted according to established clinical reference values. Gut microbiota diversity and opportunistic pathogens Alpha diversity analysis revealed that the composition of COVID-19 patients’ faecal samples was less rich (Chao index) and less diverse (Simpson's reciprocal index) compared with the control group. This resulted in reduced evenness (Simpson's evenness) and increased dominance (Dominance index) of specific bacterial species (Fig. 1A). At the community level, beta diversity analysis demonstrated that SARS-CoV-2 infection significantly altered gut microbiota composition (Fig. 1B). While common covariates such as age, sex, and BMI can influence microbiota structure, disease status was the most significant explanatory factor in this cohort. Contrary to established trends, BMI showed no significant association in our analysis. However, we identified other covariates with a significant influence on microbiota composition, including sequencing batch and age (Table S2). Therefore, we incorporated age and sequencing batch as random effects during the differential abundance GLM-based analysis. Additionally, we evaluated the potential impact of antibiotics administered to COVID-19 patients, and PERMANOVA analyses indicated that antibiotics administration did not significantly affect the overall microbiota structure within the COVID-19 group (all p > 0.18; R² < 0.02), suggesting that individual antibiotic treatments did not confound the main microbiota patterns observed. These comparisons should nevertheless be interpreted cautiously because only three patients had not received antibiotics, limiting statistical power. The absence of a detectable association may appear unexpected, given that antibiotics were prescribed much more frequently to patients with COVID-19 than to controls. However, the great variability in antibiotics prescribed and recorded in the metadata (encompassing nine distinct treatment protocols) and their distinctive targets and effects make it impossible to generalize their impact on the gut microbiota from a single within-group comparison. Although the results suggest that the recorded antibiotic regimens did not drive the primary microbiota patterns, the widespread use of antibiotics under early-pandemic clinical protocols remains an important limitation when distinguishing infection-associated dysbiosis from treatment-related effects.Fig. 1Microbiota composition of COVID-19 patients versus controls. A) Alpha diversity analysis showing boxplots with the respective distributions of Chao index, Simpson's reciprocal index, Simpson’s evenness and Dominance index for both groups. Statistical differences were evaluated using generalized linear models (GLM’s) adjusted for the different confounding variables as random effects, with p-values displayed at the top of each panel. B) Non-metric multidimensional analysis (k = 2) of the microbiota structure based on the Aitchinson distance metric. Coordinates for NMDS1 and NMDS2 dimensions are displayed in this scatterplot. C) Differential abundance for top-ranked taxonomy categories exhibiting an underrepresentation at the species level in COVID-19 patients. D) Differential abundance of top-ranked bacterial species significantly overrepresented in COVID-19 patients compared to controls. In panels C and D, data distribution are presented as boxplots, with differences assessed by GLMs and false discovery rate (FDR) p-adjusted values shown. E) Association between E. faecalis abundance and disease severity in COVID-19 patients evaluated by Wilcoxon test. A-D) N=63 COVID-19 patients and N=54 healthy controls; E) N=42 mild/asymptomatic vs. N=10 moderate, N=11 severe/critical. * p < 0.05, ** p < 0.01, *** p < 0.001. Microbiota composition of COVID-19 patients versus controls. A) Alpha diversity analysis showing boxplots with the respective distributions of Chao index, Simpson's reciprocal index, Simpson’s evenness and Dominance index for both groups. Statistical differences were evaluated using generalized linear models (GLM’s) adjusted for the different confounding variables as random effects, with p-values displayed at the top of each panel. B) Non-metric multidimensional analysis (k = 2) of the microbiota structure based on the Aitchinson distance metric. Coordinates for NMDS1 and NMDS2 dimensions are displayed in this scatterplot. C) Differential abundance for top-ranked taxonomy categories exhibiting an underrepresentation at the species level in COVID-19 patients. D) Differential abundance of top-ranked bacterial species significantly overrepresented in COVID-19 patients compared to controls. In panels C and D, data distribution are presented as boxplots, with differences assessed by GLMs and false discovery rate (FDR) p-adjusted values shown. E) Association between E. faecalis abundance and disease severity in COVID-19 patients evaluated by Wilcoxon test. A-D) N=63 COVID-19 patients and N=54 healthy controls; E) N=42 mild/asymptomatic vs. N=10 moderate, N=11 severe/critical. * p < 0.05, ** p < 0.01, *** p < 0.001. Although our microbiome molecular approach was addressed to resolve species-level meta-taxonomy, global distributions at the phylum and genus level were also outlined. Fig. S1 shows the predominant phylum and genus categories retrieved in the assessment and the significant differences found across categories. Differential abundance analysis identified over 30 bacterial species. Fig. 1C presents the species that showed a greater effect size and were more abundant in the control samples. COVID-19 patients showed large reductions in R. lituseburensis, F. saccharivorans, A. faecis, B. intestinalis, and B. uniformis, as well as increased abundance of opportunistic pathogenic bacterial species belonging to the genera Acinetobacter, Pseudomonas, Klebsiella, and Streptococcus. Notably, some species overrepresented in COVID-19 patients, such as Klebsiella pneumoniae and Streptococcus pneumoniae, are commonly associated with nosocomial infections and lung diseases. Furthermore, we observed an increased abundance of other important nosocomial pathogens, including Enterococcus faecium, E. faecalis, Klebsiella aerogenes (previously Enterobacter aerogenes), and Acinetobacter lwofii (Fig. 1D and Table S3). We further explored associations between species with differential abundance between groups and the wide range of clinical variables recorded in COVID-19 patients. E. faecalis and E. faecium were associated with dyspnoea, K. pneumoniae and K. aerogenes were negatively correlated with haemoglobin levels, and K. pneumoniae was positively correlated with Acinetobacter faecalis abundance. Disease severity, classified by clinical symptoms, was significantly associated with E. faecalis abundance (Kendall's tau = − 0.26, FDR = 0.010; Fig. 1E), indicating that asymptomatic or mildly symptomatic patients had lower levels of this bacterium (Wilcoxon test = 599, p = 0.019). Immunoregulatory role of Bacteroides uniformis To investigate how bacterial species showing different abundances (in COVID-19 patients versus the control group) might influence infection, we evaluated their effect on the inflammatory response of PBMC cultures exposed to PepMix SARS-CoV-2 (S-MUT), a pool of four peptides derived from a peptide scan (15mers with an 11 aa overlap) through the SARS-CoV-2 S glycoprotein. The immune effects of the bacteria on PBMCs were altered when they came into contact with SARS-CoV-2 antigens. Fig. 2 shows the ratio of cytokines produced by the PBMCs in the presence of the viral peptides plus each of the bacterium to those produced only by the bacterium. Of the strains tested, B. uniformis IATA25 exhibited the strongest anti-inflammatory effects, significantly reducing IFNγ and IL-6 ratios in response to SARS-CoV-2 peptides (Fig. 2A-C). No significant differences were observed for IL-12p70 and IL-5 (data not shown). Similar results were seen when calculating the ratios of the anti-inflammatory cytokine IL-4 to pro-inflammatory cytokines (IL-4/TNF, IL-4/IFNγ, IL4/IL-6). B. uniformis IATA25 also exhibited the highest IL-4/TNF, IL-4/IFNγ, and IL4/IL-6 ratios (Fig. 2D-F), highlighting its immunomodulatory potential. Meanwhile, A. faecis and R. lituseburensis decreased IL-6 levels, and F. saccharivorans induced the highest levels of TNF and IL-6, indicating its ability to elicit a pro-inflammatory response. These results suggest that there are species-specific effects on cytokine modulation in response to SARS-CoV-2 antigens, which may partly explain the associations between gut microbiota compositional changes and the disease.Fig. 2Assessment of the immune response induced by gut bacteria in PBMCs exposed to SARS-CoV-2 antigens. A-C) Ratio of the pro-inflammatory cytokines TNF (A), IFNγ (B) and IL-6 (C) produced by PBMCs in response to the viral antigens plus bacterium relative to the bacterium alone (control condition). Values > 1 indicate an increased pro-inflammatory cytokine response due to the presence of viral antigens. D-F) Relative anti-inflammatory ratios of IL-4 to each pro-inflammatory cytokine (IL-4/TNF (D), IL-4/IFNγ (E) and IL-4/IL-6 (F)) normalized to the baseline levels with the bacterium alone (values > 1 indicate a shift toward an anti-inflammatory profile). Full calculation details are provided in Material and Methods. Statistical significance across conditions was evaluated using one-way ANOVA followed by Tukey’s post-hoc test for multiple comparisons. Data are presented as mean ± SEM. * p < 0.05, ** p < 0.01. Assessment of the immune response induced by gut bacteria in PBMCs exposed to SARS-CoV-2 antigens. A-C) Ratio of the pro-inflammatory cytokines TNF (A), IFNγ (B) and IL-6 (C) produced by PBMCs in response to the viral antigens plus bacterium relative to the bacterium alone (control condition). Values > 1 indicate an increased pro-inflammatory cytokine response due to the presence of viral antigens. D-F) Relative anti-inflammatory ratios of IL-4 to each pro-inflammatory cytokine (IL-4/TNF (D), IL-4/IFNγ (E) and IL-4/IL-6 (F)) normalized to the baseline levels with the bacterium alone (values > 1 indicate a shift toward an anti-inflammatory profile). Full calculation details are provided in Material and Methods. Statistical significance across conditions was evaluated using one-way ANOVA followed by Tukey’s post-hoc test for multiple comparisons. Data are presented as mean ± SEM. * p < 0.05, ** p < 0.01. Role of B. uniformis in SARS-CoV-2 cellular entry and infection Considering the anti-inflammatory effects of B. uniformis IATA25 in response to the viral antigens, we also investigated whether this strain affected cell entry and infectivity using an in vitro cell model of SARS-CoV-2 infection. To do this, we conducted the experiments with VSV-XBB1.5 pseudovirus due to infrastructure limitations for conducting experiments with the SARS-CoV-2 virus. Therefore, we use a well-established model of SARS-CoV-2 spike mediated viral entry. We observed that the bacterium did not protect susceptible VeroE6-TMPRSS2 cells from infection with vesicular stomatitis virus (VSV) pseudotyped with the S protein of the SARS-CoV-2 XBB1.5 variant (VSV-XBB1.5). Figures 3A and 3B show the extent of infection measured by GFP fluorescence. To corroborate the immune properties of B. uniformis IATA25 toward this new viral stimulus, we also tested its ability to modulate cytokine production by PBMCs exposed to VSV-XBB1.5 or VSV pseudotyped with its native glycoprotein (VSV-G). B. uniformis IATA25 elicited a larger anti-inflammatory effect than the VSV-XBB1.5 or the control VSV-G pseudoviruses (Fig. 3C).Fig. 3Evaluation of the role of B. uniformis IATA25 on viral infection and inflammation. VSV-XBB1.5 refers to VSV pseudotyped with the SARS-CoV-2 XBB1.5 spike (S) protein, whereas VSV (VSV-G) refers to VSV pseudotyped with its native glycoprotein. A, B) Fluorescence images and quantification of VSV-XBB1.5 infection in VeroE6-TMPRSS2 cells in the presence or absence of B. uniformis IATA25. GFP expression was examined by live-cell microscopy (A) and the fold-change of GFP fluorescence in bacteria-treated samples relative to GFP fluorescence of VSV-XBB1.5 alone is shown (B). (C) Ratio of the anti-inflammatory cytokine IL-4 secreted by PBMCs co-cultured with B. uniformis IATA25 after addition of either control VSV-G or VSV-XBB1.5 pseudotype virus, mimicking SARS-CoV-2 infection. Statistical analyses were performed by one-way ANOVA followed by post hoc Tukey’s multiple comparison tests (n=4 biological replicates per group). Data are expressed as mean ± SEM. * p < 0.05, ** p < 0.01. Evaluation of the role of B. uniformis IATA25 on viral infection and inflammation. VSV-XBB1.5 refers to VSV pseudotyped with the SARS-CoV-2 XBB1.5 spike (S) protein, whereas VSV (VSV-G) refers to VSV pseudotyped with its native glycoprotein. A, B) Fluorescence images and quantification of VSV-XBB1.5 infection in VeroE6-TMPRSS2 cells in the presence or absence of B. uniformis IATA25. GFP expression was examined by live-cell microscopy (A) and the fold-change of GFP fluorescence in bacteria-treated samples relative to GFP fluorescence of VSV-XBB1.5 alone is shown (B). (C) Ratio of the anti-inflammatory cytokine IL-4 secreted by PBMCs co-cultured with B. uniformis IATA25 after addition of either control VSV-G or VSV-XBB1.5 pseudotype virus, mimicking SARS-CoV-2 infection. Statistical analyses were performed by one-way ANOVA followed by post hoc Tukey’s multiple comparison tests (n=4 biological replicates per group). Data are expressed as mean ± SEM. * p < 0.05, ** p < 0.01. SCFAs production by B. uniformis IATA25 and E. faecalis CECT481 The ability of B. uniformis IATA25 and E. faecalis CECT481 to produce SCFAs was evaluated in vitro to elucidate the mechanisms underlying their association with the infection. E. faecalis significantly increased the concentration of acetic acid in its culture medium compared to the control (Fig. 4A), but it did not generate propionic or butyric acid (data not shown). By contrast, B. uniformis IATA25 increased the concentration of propionic acid relative to control medium (Fig. 4B-D), but not that of acetic or butyric acid. When total cell counts were used to normalise SCFA production, the results confirmed that B. uniformis IATA25 produced significantly more propionic acid than E. faecalis (Fig. 4F). No difference was detected in acetic acid production between the two bacterial species (Fig. 4E), while cell-normalized butyric acid was significantly decreased by B. uniformis IATA25 (Fig. 4G).Fig. 4Quantification of SCFAs produced by B. uniformis IATA25 and E. faecalis CECT481, using gas chromatography-mass spectrometry (GC-MS). (A) Acetic acid concentration in control medium (BHI) and in the cell-free culture supernatant of E. faecalis, (B–D) Quantification of acetic acid (B), propionic acid (C), and butyric acid (D) in Schaedler (SCH) medium and in the cell-free culture supernatant of B. uniformis IATA25. (E-G) Comparison of acetic acid (E), propionic acid (F), and butyric acid (G) production between B. uniformis IATA25 and E. faecalis. The production was normalized to the total number of bacterial cells. Statistical differences were evaluated usign two-tailed Mann-Whitney U test for all comparisons, excepto for panel F, where Welch’s t-test was applied to account for unequal variance (n=8 biological replicates per group). Data are expressed as mean ± SEM. * p < 0.05, *** p < 0.001. Quantification of SCFAs produced by B. uniformis IATA25 and E. faecalis CECT481, using gas chromatography-mass spectrometry (GC-MS). (A) Acetic acid concentration in control medium (BHI) and in the cell-free culture supernatant of E. faecalis, (B–D) Quantification of acetic acid (B), propionic acid (C), and butyric acid (D) in Schaedler (SCH) medium and in the cell-free culture supernatant of B. uniformis IATA25. (E-G) Comparison of acetic acid (E), propionic acid (F), and butyric acid (G) production between B. uniformis IATA25 and E. faecalis. The production was normalized to the total number of bacterial cells. Statistical differences were evaluated usign two-tailed Mann-Whitney U test for all comparisons, excepto for panel F, where Welch’s t-test was applied to account for unequal variance (n=8 biological replicates per group). Data are expressed as mean ± SEM. * p < 0.05, *** p < 0.001. Discussion In this cross-sectional study comparing patients hospitalized with SARS-CoV-2 infection versus healthy controls, we characterized clinical, immunological, and gut microbiota alterations associated with SARS-CoV-2 infection. Our findings support that COVID-19 is not only a respiratory disease but also involves significant intestinal alterations alongside systemic inflammation. Global epidemiological surveillance reports from the WHO, indicate that SARS-CoV-2 continues to circulate worldwide with persistent waves of hospitalizations, ICU admissions, and mortality, highlighting that viral transmission and severe outcomes remain a public health concern (World Health Organization 2026). Indeed, recent surveillance data reflect notable regional surges in transmission, particularly in countries such as Thailand, Brazil, Mexico, and Colombia, which have reported marked increases in case number over recent 28-day surveillance periods (World Health Organization 2026). In this study, we combined biochemical, hematologic and metagenomic data profiling from patients with experimental functional assays of specific commensal bacteria in in vitro infection models to identify microbial signatures and their potential functional roles in the host response to SARS-CoV-2. COVID-19 patients exhibited characteristic hyperinflammation biochemical profiles, including elevated ferritin, ALT, LDH, CRP, D-dimer, and IL-6 serum levels, which are classical biomarkers of inflammation and tissue damage (Zhu et al., 2020, Safiabadi Tali et al., 2021, Ji et al., 2020, Cheng et al., 2020). While these markers align with previous reports linking hyperinflammation and cytokine dysregulation in COVID-19 (Zhu et al., 2020, Hu et al., 2021, Ragab et al., 2020, Zhang et al., 2023, de et al., 2021), they did not correlate with disease severity or worse prognosis in our cohort, likely due to the small proportion of moderate-to-severe cases. Serological testing showed that only 69.8% patients were positive for IgM and IgG antibodies, reflecting know inter-individual heterogeneity in kinetics of humoral responses, sampling time point, or underlying comorbidities (Li et al., 2020, Xu et al., 2020, Shakoor et al., 2021). We also observed a decrease in serum vitamin D levels, a factor previously associated with compromised intestinal barrier integrity and microbial imbalance, dysregulation ACE2 expression, and increased susceptibility to infection (Hashimoto et al., 2012, Shakoor et al., 2021, Zhou et al., 2021, Merzon et al., 2020, Yildiz et al., 2018, Vinogradova et al., 2025, Seibert et al., 2022). Thus, a plausible mechanistic explanation for the alterations in the gut microbiota observed in our cohort is provided by both systemic inflammation and reduced vitamin D levels. We investigated the gut microbiota’s potential as a marker of disease severity and prognosis in COVID-19. Metagenomic profiles showed alpha diversity and marked shifts in microbial community structure in COVID-19 patients, corroborating previous studies (Rossini et al., 2024, Zuo et al., 2020, Ng et al., 2023, Rabaan et al., 2021). Differential abundance analysis revealed a significant enrichment of potentially opportunistic and nosocomial pathogens, including species of the genera Klebsiella, Enterococcus, Streptococcus, and Acinetobacter (Koupaei et al., 2024, das et al., 2024, Devoe et al., 2022). Several of these taxa are recognized respiratory pathogens (e.g., K. pneumoniae). This, together with the depletion of beneficial commensals, supports the clinical relevance of the bidirectional gut-lung axis in viral respiratory diseases (Rossini et al., 2024, Ng et al., 2023, Koupaei et al., 2024, Chen et al., 2021, Zuo et al., 2020, Keely et al., 2012, Ou et al., 2023, Sencio et al., 2021, Ye et al., 2025). An imbalance in the gut microbiota may influence respiratory outcomes by promoting systemic inflammation, partly through the impairment of gut epithelial barrier integrity, and enabling the translocation of bacterial components, metabolites, and opportunistic pathogens into the systemic circulation, thereby modulating immune responses in the lung (Yeoh et al., 2021, Ragab et al., 2020, Zhang et al., 2023). Notably, the present study identified a positive correlation between E. faecalis and disease severity, probably related to the species’ pathogenicity, confirming and refining previous genus-level observations (Gaibani et al., 2021, Wang et al., 2023, das et al., 2024, Devoe et al., 2022, Ren et al., 2021. Although our full-length rRNA operon amplicon approach (16S-ITS-23S) has higher resolution than standard 16S rRNA sequencing, it does not enable direct quantification of virulence factors or antimicrobial resistance (AMR) genes. However, it is widely documented that strains of E. faecalis often carry well-characterised virulence factors that facilitate adhesion, biofilm formation, immune evasion, and tissue invasion (Devoe et al., 2022, Wu et al., 2021, Giacobbe et al., 2021, Strateva et al., 2016, Anderson et al., 2016, Filipová and Bujdáková, 2005, Rams et al., 2013, Esmail et al., 2019, Farman et al., 2019). The increased relative abundance of E. faecalis in our patients clearly reflects profound ecological alterations in their gut microbiota, characterised by increases in opportunistic pathogens with virulence and drug-resistant genes. Conversely, the abundance of beneficial commensal such as R. lituseburensis, F. saccharivorans, A. faecis, B. intestinalis, and B. uniformis was significantly reduced in COVID-19 patients (Rossini et al., 2024, Gaibani et al., 2021, Wang et al., 2023, Safiabadi Tali et al., 2021, Nobre et al., 2022, Zuo et al., 2020, Bredon et al., 2025). While some studies have observed increased levels of the genus Bacteroides in patients with SARS-CoV-2 infection (Mancabelli et al., 2022), species-level resolution is essential, as the functional properties and virulence factors within this genus vary markedly by species and strain (Nobre et al., 2022, Becker et al., 2021). Regulation of the immune response, including to infectious agents, relies on the cooperation between immunocompetent cells and the gut microbiota (Rossini et al., 2024, Schirmer et al., 2016, Troisi et al., 2021, Wittig and Zeitz, 2003, Yeoh et al., 2021). Functionally, our immunological study revealed that B. uniformis IATA25 exhibits significant anti-inflammatory effects against SARS-CoV-2 antigens, which may help reduce virus-induced hyperinflammation. These immunomodulatory properties may be mediated by bacterial surface structural cell components, such as Polysaccharide A (Carasso et al., 2024), or microbiota-derived metabolites, including SCFAs, which are key mediators of host-microbiota immune interactions (Ratajczak et al., 2019, Akhtar et al., 2021). Consistent with this, we found that our COVID-19 patients had reduced abundance of SCFA-producing bacteria, such as F. saccharivorans and A. faecis (which produce butyrate) (Ratajczak et al., 2019, Akhtar et al., 2021, Pihelgas et al., 2024, Singh et al., 2022), and B. intestinalis and B. uniformis (which produce acetate and propionate) (Ng et al., 2023, Gaibani et al., 2021, Gu et al., 2020, Reinold et al., 2021, Liu et al., 2025). Experimentally, we demonstrated that B. uniformis IATA25 produces substantial amounts of propionic and acetic acids. Propionate can directly modulate inflammatory signalling and support cross-feeding interactions that fuel butyrate production by other commensals, thereby potentially protecting against COVID-19 (López-Almela et al., 2021, Fusco et al., 2023, Zhang et al., 2023). In turn, butyrate is known for its anti-inflammatory properties, its ability to reinforce mucosal barrier integrity across the respiratory and intestinal tracts, and its ability to downregulate ACE2 expression in the gut (López-Almela et al., 2021, Vergara and Sánchez, 2017, Nabizadeh et al., 2023, Włodarczyk et al., 2022, Souza et al., 2004;). Although our study was not designed to establish a direct causal relationship between SCFA production and the immunomodulatory effects of B. uniformis IATA25, its ability to produce propionate provides a biologically plausible mechanism that may contribute to the anti-inflammatory profile observed in our PBMC assays. By contrast, our results suggest that commensal bacteria such as B. uniformis may indirectly hinder viral entry into host cells (Ng et al., 2023, Wang et al., 2023), through coordinated action with butyrate producers. Moreover, the fermentative ability of B. uniformis, linked to the generation of other SCFAs and propionate, could also contribute to maintaining immune homeostasis and establishing a beneficial gut ecosystem that prevents colonisation by opportunistic pathogens that may aggravate SARS-CoV-2 infection. Despite our interesting findings, this study has several limitations. First, the cross-sectional design precludes establishing direct causality, and the relatively modest cohort size – predominantly composed of mild infections – impedes the ability to reach the statistical power necessary to detect microbiota signatures associated with more severe disease outcomes. Second, while PERMANOVA indicated that antibiotic administration did not account for significant community variance, the small number of untreated patients (n=3) prevents a definitive statistical exclusion of antibiotics as a confounding factor, an inherent challenge in cohorts from the first pandemic wave. Third, we did not profile functional metabolomics or non-bacterial kingdoms (virome, mycobiome) in patient biosamples, and we limited SCFA quantification to bacterial monoculture supernatants since in PBMCS we used UV-inactivated bacteria. Finally, information on dietary habits and lifestyle could not be collected because of the critical conditions existing during the first wave of the pandemic, when hospitals were overwhelmed. Nevertheless, most participants were residents of the same geographic area of Spain and typically adhered to a Mediterranean diet, which may have reduced the potential impact of this variable on the results. In summary, this work presents an in-depth exploratory study of gut microbiota signatures in COVID-19 patients. The results obtained through long-amplicon DNA sequencing, support robust associations between opportunistic bacterial pathogens and disease severity. We also identified commensal bacterial species and strains, such as B. uniformis IATA25, as potential immunomodulators that may reduce complications of SARS-CoV-2 infection. Our finding suggests that B. uniformis could be considered a biomarker of a healthy gut microbiota in the context of SARS-CoV-2 infection, with immunoregulatory potential, although further functional studies are warranted. Funding This study was supported by CSIC grant BICOIN (Ref. 202040E227) and the European Commission—NextGeneration EU funds through CSIC Interdisciplinary Thematic Platform Global Health+ (Ref. SGL2103028). The Accreditation as Center of Excellence Severo Ochoa CEX2021-001189-S funded by MICIU/AEI/10.13039/501100011033 is also fully acknowledged. Project/Equipment funded by Consellería de Sanitat Universal i Salut Pública (Generalitat Valenciana, Spain and the EU Operational Program of the European Regional Development Fund (ERDF) for the Valencian Community 2014-2020, within the framework of the REACT-EU programme, as the Union’s response to the COVID-19 pandemic. Grant CNS2022-135100, funded by MCIN/AEI/10.13039/501100011033 and by the European Union NextGenerationEU/PRTR. CIB-CSIC (SGL2103015) and Spanish Ministry of Science project (PID2023-150693OB-I00). Informed Consent Statement Written informed consent has been obtained from the patient(s) to publish this paper Author contribution ConceptualizationYS, MLSM, RG, MM, EMMRFormal analysisVTE, CFC, ABPInvestigation (experimental performance)VTE, AIAA, LMG, EMMR, CFCData CurationVTE, MLSM, CFCResourcesYS, JGN, ALC, MLSM, GVMMWriting - Original DraftVTE, MLSM, CFC, ABPWriting - Review & EditingYS, EMMR, VTE, RG, MM, MLSMSupervisionYS, EMMR ConceptualizationYS, MLSM, RG, MM, EMMRFormal analysisVTE, CFC, ABPInvestigation (experimental performance)VTE, AIAA, LMG, EMMR, CFCData CurationVTE, MLSM, CFCResourcesYS, JGN, ALC, MLSM, GVMMWriting - Original DraftVTE, MLSM, CFC, ABPWriting - Review & EditingYS, EMMR, VTE, RG, MM, MLSMSupervisionYS, EMMR Declaration of competing 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. All the authors declare no conflict of interest.