Ozone water enema activates SIRT1-Nrf2/HO-1 pathway to ameliorate gut dysbiosis in mice receiving COVID-19 patient-derived faecal microbiota

This study followed the principles outlined in the Declaration of Helsinki, and written informed consent was obtained from all participants. Approval for the study was granted by the Medical Ethics Committee of Shanghai East Hospital (no. 2022202). Between December 2022 and January 2023, we enrolled 21 patients with COVID-19 from the Respiratory Critical Care Emergency Department of Shanghai East Hospital. Inclusion criteria for stool sample collection included a positive test for coronavirus antigen, the presence of at least four respiratory complications (e.g. fever, persistent cough, shortness of breath, respiratory tract infections and pneumonia), SaO2≤90% or PaO2≤60 mmHg, no haemolysis, body mass index between 18.5 and 24, aspartate transaminase (AST) and alanine aminotransferase (ALT) values≥20 mmHg and no haemorrhagic fever. Faecal samples were collected in the hospital, mixed with cold PBS (4 °C) at a 1 : 3 ratio. Samples were refrigerated at 4 °C before faecal microbiota transplantation (FMT), and the trial’s exclusion criteria included underlying respiratory diseases (e.g. chronic obstructive pulmonary disease), chronic gastroenteritis, history of gastrointestinal surgery, psychiatric disorders, Parkinson’s disease, diabetes and pregnancy.

Specific pathogen-free male C57BL/6J mice (16 weeks old, 25–30 g; Shanghai Centre for Model Organisms) were housed under a 12 h light–dark cycle with ad libitum access to food and water. The mice (n=10/group) were randomly assigned to three experimental cohorts:

Sham: oral gavage with PBS on days 1–7, followed by double-distilled water (ddH₂O) enema on days 8–14.

FMT: FMT via oral gavage on days 1–7 [3335], then ddH2O enema on days 8–14.

FMT+O₃: FMT identical to group 2 during days 1–7, followed by ozonated water (30 µg ml⁻¹ [3638]) enema on days 8–14.

FMT preparation: fresh faecal samples from 21 critically ill COVID-19 patients were pooled into seven donor groups assigned to days 1–7 (3 patients/group). Then, on days 1–7 of the experiment, faeces from the group of the day (3 patients) were collected, mixed well and homogenized in PBS at a ratio of 1 : 3 (w/v), vortexed for 5 min and then left to stand (30 min, 4 °C). The supernatant was filtered through a 70 µm cell strainer and administered (100 µl/mouse) within 2 h post-preparation.

Ozonated water administration: on days 8–14, FMT+O₃ mice were anesthetized with 3% sevoflurane (10 min exposure), followed by insertion of a polyethylene catheter (4 cm depth from anal verge). Ozonated water (200 µl/mouse) was infused slowly using a 1 ml syringe. To ensure uniform colonic distribution, mice were maintained in a vertical inverted position for 10 min post-administration before returning to standard housing.

Terminal sampling: all mice underwent terminal blood/tissue collection under sevoflurane anaesthesia at day 14, 8 h after the final intervention. After sacrifice, tissues were rapidly excised using sterile surgical instruments. The collected tissues were immediately placed in pre-chilled collection tubes containing appropriate buffers or fixatives to prevent degradation and maintain sample integrity. For RNA isolation, tissues were snap-frozen in liquid nitrogen and stored at −80 °C until processing with TRIzol. For histological analysis, tissues were fixed in 10% neutral buffered formalin immediately after collection, processed through routine paraffin embedding, sectioned and then subjected to staining procedures as described.

Randomization protocol: mice were allocated to treatment groups (Sham, FMT and FMT+O₃) using a stratified randomization approach to ensure baseline homogeneity:

Step 1: All mice (n=30) were initially grouped by body weight (25–30 g range) into three strata (25–26.5, 26.6–28.5 and 28.6–30 g).

Step 2: Within each stratum, mice were randomly assigned to the three treatment groups using a computer-generated random number sequence (GraphPad Prism v9.0).

Step 3: Final group assignments were validated to ensure no significant differences in baseline weight (one-way ANOVA, P=0.76) or age (Kruskal–Wallis test, P=0.82).

Blinding during outcome assessment: to minimize observer bias, a double-blind design was employed:

Intervention phase: personnel administering FMT or ozonated water enemas were unaware of group assignments. Treatments were coded as ‘A’, ‘B’ and ‘C’ (decoded post-analysis).

Endpoint measurements: objective parameters (e.g. serum cytokines, faecal 16S sequencing and oxidative stress markers) and semi-quantitative assessments (e.g. histopathology scoring).

Total RNAs were isolated from tissues using TRIzol reagent (Thermo Fisher Scientific, USA), and cDNA was synthesized using the ChamQ Universal SYBR qPCR Master Mix (Vazyme, China). The qPCR procedure involved an initial 30 min incubation at 95 °C, followed by 40 cycles of 10 s at 95 °C and 30 s at 60 °C. SYBR Green System (Yeasen, Shanghai) was utilized for qPCR according to the provided instructions. The relative gene expression was calculated using the 2−ΔΔCt method, with GAPDH serving as an internal control and reference gene. The primers used in the study are listed in Table 1.

Sampled tissues were fixed in 10% neutral formalin for 24 h and then dehydrated progressively in graded alcohols, with 1 h of immersion at each concentration, followed by hyalinization in xylene before paraffin embedding using an embedding machine. Three consecutive sections of each sample were stained, and the whole sections were analysed to ensure a thorough assessment and to obtain a representative and complete picture of the staining pattern of the whole tissue. Tissue paraffin sections with a thickness of 5 µm were first placed in an oven at 65 °C for 45 min, then deparaffinized and rehydrated. Antigen retrieval was achieved by heating sections in citrate buffer (pH 6.0) using an autoclave. Tissue sections were then incubated in 3% H₂O₂ for 15 min, followed by a PBS rinse (5 min, 3 times) and then incubated with 5% BSA at 25 °C for 30 min. The sections were then incubated with monoclonal antibodies against CD45 (1 : 500, Cell Signaling), CD11b (1 : 700, Cell Signaling) [39], F4/80 (1 : 1,000, Cell Signaling) [39] and NLRP3 (1 : 700, Immunoway) at 4 °C overnight. After rinsing with PBS (5 min, 3 times), the sections were incubated with polymerized horseradish peroxidase-labelled goat anti-rabbit IgG (1 : 1,000, Beyotime) for 40 min at 37 °C. Detection was performed by DAB staining (Vector Labs), and sections were counterstained with haematoxylin. Finally, the sections were dehydrated, cleared and fixed with neutral gel.

Positive and negative controls were used in each experiment to confirm staining specificity. For negative controls, sections were incubated with isotype-specific immunoglobulins at the same protein concentration as the primary antibody, ensuring that any observed staining was due to specific binding rather than nonspecific interactions [40].

After the collected tissue samples were immediately embedded in Tissue-Tek, using a cryotome, the tissues were cut into 6-µm-thick sections and stored at −80 °C. Slides were air-dried for 30 min at room temperature followed by 4% polyfluoroalkoxy fixation. Next, tissue samples were blocked for 1 h at room temperature in a humidity chamber containing 5% goat serum. After blocking, samples were incubated with the primary antibody zonula occludens-1 (ZO-1) (1 : 200, Abcam) and left in a humidity chamber at 4 °C overnight. Samples were washed thoroughly in PBS and then incubated with the secondary antibodies for 1 h in a humidity chamber. After incubation, sections were thoroughly washed again in PBS. Finally, the sections were mounted in an aqueous DAPI medium (Life-iLab) to counterstain nuclei. Mucus staining was performed according to an established protocol [41]. Briefly, colon tissue sections containing faeces were fixed using the Carnoy fixation method (60% absolute methanol, 30% chloroform and 10% glacial acetic acid). After paraffin embedding, mucus and intestinal bacteria were stained with an anti-Mucin-2 primary antibody (1 : 1,000, Santa Cruz).

The total proteins in the tissues were first extracted using RIPA lysis buffer (Epizyme Biotech, China) supplemented with a 1% protease inhibitor cocktail (Epizyme Biotech, China). The protein concentration was determined by centrifugation at 4 °C and 12,000 r.p.m. (16,000 g) for 15 min, and the bicinchoninic acid assay kit (Solarbio, China) was used for detection. Subsequently, the samples were boiled at 100 °C for 5 min, and then, the proteins were separated on a 12.5% gel prepared using the SDS-PAGE Gel Fast Preparation Kit (Epizyme Biotech, China) for electrophoretic separation and immunodetection. Electrophoresis was followed by transfer to a PVDF membrane (Millipore, Burlington, MA, USA) at a constant current of 0.25 A for 90 min. The membranes were blocked with 5% nonfat milk-TBST solution for 1 h and then incubated overnight at 4 °C with the following primary antibodies: rabbit anti-GAPDH (1 : 5,000, Proteintech), rabbit anti-P-P-65 NF-κB (1 : 1,000, Wanleibio), rabbit anti-P-65 NF-κB (1 : 1,000, Wanleibio), rabbit anti-caspase-1 (1 : 1,000, Proteintech), rabbit anti-interleukin-1β (IL-1β; 1 : 1,000, Abcam), rabbit anti-Nrf2 (1 : 200, Santa Cruz) and rabbit anti-SIRT1 (1 : 1,000, Proteintech). Protein bands were detected with HRP-conjugated rabbit secondary antibody (1 : 5,000, Beyotime). After incubation with the secondary antibody, specific bands were visualized using chemiluminescence imaging on a ChemiDoc XRS+ System (Bio-Rad, USA) with an ECL kit (Juhemei, China). Protein bands were analysed using ImageJ software.

On day 14 of the experiment, faecal samples were collected from 21 mice. Total genomic DNA was extracted from these samples using the CTAB method. The concentration and purity of the DNA were assessed on 1% agarose gels, which are ideal for separating DNA fragments ranging from 500 bp to 10 kb. Based on these assessments, the DNA was diluted to a concentration of 1 ng µl⁻¹ with sterile water. For the amplification of the 16S rRNA genes (16S V3–V4 regions), specific primers (341F: 5′-CCTAYGGGRBGCASCAG-3′ and 806R: 5′-GGACTACNNGGGTATCTAAT-3′) with barcodes were used. Each PCR reaction was performed using 15 µl of Phusion^® High-Fidelity PCR Master Mix (New England Biolabs), 2 µM of both forward and reverse primers and ~10 ng of template DNA. The thermal cycling conditions consisted of an initial denaturation at 98 °C for 1 min, followed by 30 cycles of denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s and elongation at 72 °C for 30 s. A final extension step was performed at 72 °C for 5 min. To analyse the PCR products, equal volumes of 1× TAE buffer were mixed with the PCR products, and electrophoresis was performed on 2% agarose gels, which provide better resolution for smaller DNA fragments due to their smaller pore size. The PCR products were then mixed in equidensity ratios. The mixture of these fragmented PCR products was purified using the Universal DNA Purification Kit (Tiangen, China).

Sequencing libraries were generated using the NEBNext^® Ultra DNA Library Prep Kit (Illumina, USA) following the manufacturer’s protocol, with index codes added for sample identification. The quality of the libraries was assessed using the Agilent 5400 system (Agilent Technologies Co. Ltd., USA) to ensure proper fragment size distribution and concentration. The validated libraries were then sequenced on the Illumina platform, generating 250 bp paired-end reads for subsequent bioinformatics analysis.

The analysis was conducted following the ‘Atacama soil microbiome tutorial’ by Qiime2docs along with customized programme scripts (https://docs.qiime2.org/2019.1/↗). Briefly, the raw data FASTQ files were imported into a format compatible with the QIIME2 system using the qiime tools import programme. Demultiplexed sequences from each sample were quality filtered, trimmed, denoised and merged, and then, the chimeric sequences were identified and removed using the QIIME2-Dada2 plugin to obtain the feature table of amplicon sequence variants (ASVs). The QIIME2 feature-classifier plugin was used to align ASV sequences to a pre-trained GREENGENES 13_8 99% database (trimmed to the V3–V4 region bound by the 338F/806R primer pair) and generate the taxonomy table. Contaminating mitochondrial and chloroplast sequences were filtered using the QIIME2 feature-table plugin. Various statistical methods, including ANCOM, ANOVA, Kruskal–Wallis, linear discriminant analysis effect size (LEfSe) and DESeq2, were employed to identify bacteria with different abundances among samples and groups. Diversity metrics were calculated using the core-diversity plugin within QIIME2. Alpha diversity indices at the feature level, such as observed OTUs (Operational Taxonomic Units), the Chao1 richness estimator, the Shannon diversity index and Faith’s phylogenetic diversity index, were calculated to estimate the microbial diversity within each sample. Beta diversity distance measurements, including Bray–Curtis, unweighted UniFrac and weighted UniFrac, were performed to investigate the structural variation of microbial communities across samples, and the results were visualized using principal coordinate analysis (PCoA) and nonmetric multidimensional scaling. Partial least squares discriminant analysis was also used as a supervised model to reveal the variation of microbiota among groups, employing the ‘plsda’ function in the R package ‘mixOmics’. Redundancy analysis was performed to explore the association of microbial communities with environmental factors based on the relative abundance of microbial species at different taxonomic levels using the R package ‘vegan’. Co-occurrence analysis was performed by calculating Spearman’s rank correlations between dominant taxa, and the network diagram was used to display the associations among taxa. Additionally, the potential functional profiles of microbial communities based on KEGG Orthologs were predicted using PICRUSt. Unless otherwise stated, standard parameters were used in the analysis.

For comparisons of two groups, significance was tested by an unpaired two-tailed Student’s t-test after assessing normality using the Shapiro–Wilk test. For more than two groups, we first checked the homogeneity of variances using Levene’s test. If variances were equal, we used a one-way ANOVA followed by a Tukey test with an adjusted P-value for multiple comparisons. If variances were unequal, Welch’s ANOVA was employed instead. For non-normally distributed data, two groups were compared using the Wilcoxon–Mann–Whitney test, and for more than two groups, the Kruskal–Wallis test was used with the Dunn–Bonferroni test (GraphPad Prism 8). The data between experimental groups were considered significant as *P<0.05. Effect sizes (Cohen’s d and η²) were calculated to provide context to the magnitude of observed effects.

In an effort to understand the impact of SARS-CoV-2 on the gut microbiota, we conducted a study involving the transplantation of faecal flora from COVID-19 patients to mice. A total of 21 COVID-19 patients were recruited from the Respiratory Critical Care Emergency Department of Shanghai Oriental Hospital. The criteria for inclusion and exclusion of faecal sample collection are outlined in Table S1. To assess the effects of FMT on the intestinal microbiota of mice, faecal samples were collected from three groups of eight mice each, following a 7-day intervention with ozone water enemas (Fig. 1a, b). The samples were then subjected to 16S rRNA gene sequencing. The PCoA revealed a significant shift in the intestinal microbial profile as a result of FMT (Fig. 1c). A heatmap analysis of the relative abundance at the family level showed that the FMT group had higher levels of Mycoplasmataceae, Desulfovibrionaceae, Streptococcaceae and Prevotellaceae compared to the Sham group (Fig. 1d). Notably, the FMT group exhibited lower abundance of Adlercreutzia, Lactobacillaceae, Bacteroidaceae and Anaeroplasma compared to the Sham group (Fig. 1d). In contrast, the FMT+O₃ group showed higher proportions of Adlercreutzia, Lactobacillaceae and Anaeroplasma compared to the Sham group. Interestingly, the FMT group displayed a reduction in alpha diversity, suggesting that the introduction of COVID-19 patients' gut microbiota into mice leads to the depletion of specific bacterial taxa. Further evaluation in the FMT+O₃ group revealed an increase in both the Chao1 and Shannon indices, although these remained lower than those observed in the Sham group (Fig. 1e, f). This finding suggests that ozone water enemas may potentially enhance the abundance of bacterial microbiota in mice. At the phylum level, the FMT group showed an increased relative abundance of Firmicutes, Tenericutes, Actinobacteria and Deferribacteres compared to the Sham group. Conversely, the presence of Bacteroidota was reduced in the FMT group. Additionally, the ozonated water enema treatment group exhibited an increased relative abundance of the genera Prevotellaceae_Prevotella and Bacteroides, while the abundance of Mycoplasma and Lactobacillus decreased (Fig. 1g, h). The three main groups of changing flora and their biological significance are shown in Table 2. To further explore the differences in the gut microbiota among the groups, we conducted a LEfSe analysis. This analysis revealed significant differences in the relative abundance of gut microbiomes among the three groups (Fig. 1i). These findings underscore the potential of ozone water enemas as a therapeutic intervention to modulate the gut microbiota in mice, which could have implications for the management of systemic symptoms in COVID-19 patients.

The primary objective of our study was to investigate the impact of microbiota associated with COVID-19 patients on the outcomes of WT mice following FMT and to conduct a comprehensive analysis of gut phenotypes. Our observations revealed that the mice experienced weight loss after undergoing FMT, a condition that improved with subsequent ozone water enema treatment (Fig. 2a). On day 4, a statistically significant difference in mean body weight was observed between the Sham and FMT groups (P=0.0414). By day 8, a significant difference in mean body weight was also noted between the FMT and FMT+O₃ groups (P=0.0228). However, on day 12, there was no statistically significant difference in body weight between the Sham and FMT+O₃ groups. This indicates that ozonated water treatment enhanced the body functions of the mice and reversed the weight loss caused by FMT. Furthermore, we found that the intestinal length of mice in the FMT group was shorter during the sampling period, suggesting that the transplantation of faecal microbiota from COVID-19 patients may lead to decreased appetite, compromised immunity and intestinal atrophy in the mice (Fig. 2b). Experimental data from Abdul Rashid et al. highlight the critical role of the gut microbiota in maintaining the integrity of the intestinal barrier, yet its specific impact on COVID-19 patients remains to be elucidated. Severe COVID-19 often results in multi-organ failure, which can severely impact the prognosis of the disease [4243]. To evaluate the effects of SARS-CoV-2 on the intestine, we examined the intestinal epithelium of mice after cecum transplantation from COVID-19 patients. Immunofluorescence staining of tissue sections showed a significant decrease in Muc2 protein content in the ileal epithelium of mice in the FMT group compared to the Sham group, as well as a significant decrease in the content of the tight junction protein, ZO-1 (Fig. 2c, d, f, g). Additionally, we measured the mRNA content of Muc2 and ZO-1 in the colon tissues of the three groups using real-time PCR (RT-PCR), and the results showed a significant decrease in the FMT group compared to both the Sham and FMT+O₃ groups (Fig. 2e, h). These findings suggest that the gut microbiota of COVID-19 patients adversely affects the intestinal health of mice, leading to disrupted intestinal barrier function and impaired mucus barrier, which may contribute to systemic complications. The improvements observed with ozone water enema treatment indicate its potential as a therapeutic intervention to mitigate these effects.

After undergoing FMT, mice demonstrated a markedly intensified inflammatory response, with the recruitment of inflammatory monocytes playing a pivotal role in intestinal injury. Immunohistochemical (IHC) staining for CD11b and F4/80 revealed a significant twofold increase in F4/80 cell infiltration in the colon of mice post-FMT compared to the Sham group (Fig. 3a, c). Moreover, IHC staining for CD45 indicated elevated leucocyte infiltration in the gut of mice, forming dense clusters (Fig. 3e). Compared to the Sham group, FMT mice exhibited significantly heightened inflammatory infiltration in the intestine, characterized by shorter villi, apical rupture and infiltration of inflammatory cells in the muscular propria, along with the presence of giant pannus nodules (Fig. 3e). Given the observed increase in leucocyte infiltration in the FMT group of mice, our subsequent focus was directed towards investigating gut changes to elucidate the functional mechanisms underlying the amplified systemic inflammatory response in these mice. However, following the notable reduction in macrophage and leucocyte infiltration in the FMT+O₃ group of mice compared to the FMT group, along with a tendency towards lower CD11b, F4/80 and CD45 expression (Fig. 3b, d, f), our focus shifted to understanding the mechanisms by which ozonated water enemas exert anti-inflammatory effects, leading to the resolution of inflammation in the FMT group. This suggests that ozonated water enemas may help mitigate the heightened systemic inflammatory response observed in FMT mice by modulating intestinal inflammation and immune cell infiltration. After IHC staining for the NLRP3 inflammatory factor, we observed an intriguing result. There was a significant increase in NLRP3 expression in colon tissue sections in the FMT group compared to the Sham group (Fig. 3g). Additionally, there was a significant tendency towards decreased NLRP3 expression in colon tissue after administration of ozone water enema treatment (Fig. 3h). These findings suggest that the gut microbiota from COVID-19 patients can trigger a robust inflammatory response in mice, potentially contributing to intestinal injury. The reduction in inflammatory markers and NLRP3 expression following ozone water enema treatment indicates its potential as a therapeutic intervention to mitigate these inflammatory effects.

To further evaluate the overall health of mice following FMT from critically ill COVID-19 patients, we conducted a comprehensive examination of the liver, lungs and blood across three groups of mice. IHC staining for CD45 revealed an increase in leucocyte infiltration in the liver post-FMT, accompanied by enhanced recruitment of inflammatory monocytes, which contribute to liver damage (Fig. 4a, c). In the preceding conclusion, we have already demonstrated that FMT-induced gut microbiota dysbiosis and gut epithelial barrier disruption lead to an inflammatory response. Similar inflammatory responses were observed in the lungs, as evidenced by IHC staining of lung tissue sections for CD11b (Fig. 4b, e). On day 14 of the experiment, mice that received FMT from critically ill COVID-19 patients exhibited severe pulmonary fibrosis and increased intra-alveolar exudation. RT-PCR analysis of liver and lung tissues collected on day 14 revealed significantly higher mRNA levels of CD45 in liver tissues and CD11b in lung tissues in the FMT group compared to the Sham group (Fig. 4d, f). Ozone water enema intervention reversed this inflammatory trend in both organs when compared to the FMT group. Haematological examination of blood samples collected on day 14 showed a significant increase in AST and ALT levels in FMT mice, indicating liver inflammation (Fig. 4g, h). Additionally, the expression of pro-inflammatory cytokines IL-6, IL-1β and TNF-α was significantly upregulated in the FMT group compared to the Sham group and decreased following ozone water enema treatment (Fig. 4i–k). These results suggest that FMT from critically ill COVID-19 patients can induce systemic inflammation affecting multiple organs, including the liver and lungs. The increase in liver enzymes and pro-inflammatory cytokines indicates liver and systemic inflammation. Ozone water enema treatment appears to mitigate these inflammatory effects, suggesting its potential as a therapeutic intervention to improve overall health post-FMT.

Previous experiments have showed that transplanting faecal microbiota from critically ill COVID-19 patients into mice alters the gut microbial community, potentially reducing microbiota numbers and triggering a significant systemic inflammatory response. In terms of protein expression in colonic tissues, levels of SIRT1, Nrf2, NF-κB, caspase-1 and IL-1β were measured. The protein levels of SIRT1 and Nrf2 were found to be higher in the FMT group compared to the Sham group, and these levels continued to increase even after treatment with ozone water enemas (Fig. 5a, b, d). In both the FMT group and the FMT+O₃ group, protein levels in colonic tissues showed a decrease in the expression of NF-κB, cleaved caspase-1 and IL-1β (Fig. 5a, e–g). Numerous studies have indicated that repeated enemas with low-dose ozonated water can activate the Nrf2 pathway. This activation leads to an increase in the expression of various antioxidant enzymes, such as glutathione peroxidase and catalase, which can help reduce tissue damage and organ dysfunction caused by oxidative stress [4445]. In the FMT+O₃ group, the expression of HO-1 mRNA in the intestinal tissues of mice was significantly higher compared to the FMT group (Fig. 5c). At the protein level in colonic tissues, the expression of HO-1 was also significantly increased in the FMT+O₃ group compared to the FMT group. These findings suggest that ozonated water enemas may enhance the expression of antioxidant enzymes and reduce inflammation, potentially mitigating the adverse effects of COVID-19-associated microbiota on the gut and systemic health.