Microbiota-derived extracellular vesicles mediate gut-brain axis dysfunction in long COVID

Microbiota-derived extracellular vesicles mediate gut-brain axis dysfunction in long COVID | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Biological Sciences - Article Microbiota-derived extracellular vesicles mediate gut-brain axis dysfunction in long COVID Matheus Aranguren, Kim Doyon-Laliberté, Idia Boncheva, Alexandre Villard, and 17 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8876425/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Post COVID-19 condition (Long COVID, LC) is a heterogenous post-infectious condition frequently accompanied by persistent neurological symptoms, but its mechanisms remain unclear1-4. Here we identify gut microbiota-derived extracellular vesicles (GMEVs) as effectors linking LC-associated intestinal dysbiosis to systemic and neuroinflammation. In a longitudinal, deeply phenotyped cohort, individuals with LC and neurological symptoms (LC-Neuro) show a distinct intestinal microbiome profile that persists over time. Transplantation of LC-Neuro microbiota into germ-free mice disrupts intestinal barrier integrity and induces neurobehavioral alterations and neuroinflammation. GMEVs isolated from individuals with LC activate intestinal epithelial cells, macrophages and induced pluripotent stem cell-derived microglia in vitro, engaging inflammasome signalling, impairing epithelial barrier function and promoting inflammatory cytokine production. These effects are strongest for LC-Neuro-derived GMEVs in gut epithelium and macrophage models, whereas microglial activation is observed across LC-derived GMEVs. Oral administration of LC-Neuro GMEVs to conventional mice is sufficient to induce intestinal inflammation and systemic immune activation, accompanied by neurobehavioral changes and neuroinflammation. Together, these findings implicate GMEVs as mediators of gut-brain axis dysfunction and provide a mechanistic framework linking intestinal dysbiosis to neurological sequelae in LC. Biological sciences/Microbiology/Microbial communities/Microbiome Biological sciences/Immunology/Inflammation/Chronic inflammation Health sciences/Diseases/Infectious diseases/Viral infection Biological sciences/Immunology/Mucosal immunology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Approximately 5-19% of individuals infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) develop post COVID-19 condition (PCC), also referred to as post-acute sequelae of COVID-19 (PASC) or Long COVID (LC) 2,5 . LC is a post-infectious chronic condition present for at least 3 months after acute SARS-CoV-2 infection, and it may follow a continuous, relapsing-remitting or progressive disease course affecting one or multiple organ systems 2,6 . Although more than 200 symptoms have been attributed to LC 7,8 , some of the most prevalent and debilitating manifestations include fatigue, post-exertional malaise, cognitive impairment (“brain fog”), memory and concentration difficulties, dyspnoea and cardiac dysautonomia, including postural orthostatic tachycardia syndrome 4,9,10 . These sequelae can substantially impair functional capacity and quality of life for years after infection 2,11 . Despite its growing public health burden, LC currently lacks validated diagnostic biomarkers and targeted pharmacological therapies. Accumulating evidence implicates immune dysregulation 12-14 , chronic systemic inflammation 15,16 , microclotting 17 , neuroinflammation 18,19 , autoimmunity 20,21 and viral persistence 22,23 , including prolonged viral shedding in the gastrointestinal (GI) tract 24-26 , as features of LC pathophysiology. Acute SARS-CoV-2 infection is associated with intestinal dysbiosis, particularly in individuals with severe disease 27 , and alterations in the gut microbiota can persist for at least 6 months in individuals with LC 28 . However, the mechanisms by which SARS-CoV-2-associated dysbiosis contributes to immune dysregulation and end-organ injury, including neuroinflammation, remain poorly defined. The intestinal microbiota plays a central role in immune homeostasis 29 , particularly at the mucosal barrier 30 , and dysbiosis has been linked to increased intestinal inflammation and epithelial permeability (“leaky gut”), permitting the translocation of microbial products into systemic circulation 31 . Increased intestinal permeability has been documented during acute COVID-19 infection 32,33 , and circulating markers of microbial translocation are elevated in individuals with LC 34 . Although the gut-brain axis has been implicated in neuropsychological and neurodegenerative disorders 35 , including in chronic viral infections such as human immunodeficiency virus (HIV) 36-38 , a causal relationship between intestinal dysbiosis, barrier dysfunction, immune activation, and neuroinflammation has not been demonstrated in LC. Bacteria communicate with each other and their host through the release of extracellular vesicles (EVs), nanoscale (20-400 nm) lipid bilayer particles that transport proteins, lipids, metabolites and nucleic acids 39 . EVs produced by the intestinal microbiota, here termed gut microbiota-derived extracellular vesicles (GMEVs), can cross the intestinal barrier, enter the bloodstream and act on distant tissues, thereby contributing to diverse inflammatory and metabolic disorders, including metabolic syndrome and liver disease 40-42 . We hypothesized that GMEVs produced by the perturbed microbiota of individuals with LC promote intestinal barrier dysfunction, systemic inflammation and downstream neuroinflammation. Here, we functionally characterized the faecal microbiome of individuals with LC at 3-6 and 12 months post-infection and examined the biological effects of LC-derived GMEVs using complementary in vitro and in vivo models. Our findings identify GMEVs as mechanistic mediators linking intestinal dysbiosis to barrier dysfunction, immune activation, and neuropsychological manifestations in LC. Results Women with LC and neurological symptoms have intestinal dysbiosis that transfers neurobehavioral alteration and neuroinflammation to gnotobiotic mice We analysed a sub-cohort of 103 participants from the Institut de recherches cliniques de Montréal (IRCM) Post-COVID-19 (IPCO) cohort (NCT04736732), comprising 12 pandemic controls (PC; individuals who never reported COVID-19 symptoms or tested positive for SARS-CoV-2) and 91 individuals with LC who were not hospitalized during acute infection, all of whom provided stool samples for shotgun metagenomic sequencing (Fig.1a). The median age (IQR) of PC and LC participants was 50 (46.5-63) and 47 (39-58) years, respectively, and 41.7% of PC and 64.8% of LC participants self-identified as female. Female sex was associated with higher odds of LC symptom reporting overall (odds ratio (OR)=1.26, p=6.3 x 10 -9 ; Fig.1b) including several neurological symptoms, such as trouble with memory (OR=12.2, P =0.018), confusion (OR=91.95, P =0.043), vertigo or dizziness (OR=12.44, P =0.042) and nausea (OR=2183, P =0.022). Across sexes, LC participants exhibited intestinal microbiome profiles that were distinct from PC participants at 3-6 months and 12 months following acute infection, with differences in alpha diversity, beta diversity and taxonomic features that persisted over time (Extended Data Fig.1). To assess whether neurological symptom burden tracked with specific microbiome features, LC participants were stratified by neurological symptom burden. Female LC participants were categorized as having severe neurological symptoms (Severe Neuro; ³3 of 8 assessed neurocognitive /neuropsychological symptoms), mild neurological symptoms (Mild Neuro; <3 symptoms), or no neurological symptoms (No-Neuro) (Fig.1b). In females, Severe Neuro LC participants showed higher microbial diversity and evenness than No Neuro LC participants (Extended Data Fig.2a), together with shifts in community composition (Extended Data Fig.2b). A similar pattern was observed in males, with neurological symptom burden associated with differences in diversity and beta diversity at 6 and/or 12 months (Extended Data Fig.2c,d). Factor analysis of mixed data (FAMD) further supported an association of neurological and respiratory symptom dimensions with microbiome composition in both sexes (Extended Data Fig.3). At the phylum level, community composition differed across neurological symptom strata and timepoints in both females and males (Fig.1c). Taxonomic analyses identified sex-specific patterns of dysbiosis associated with neurological symptom burden. In females, Severe Neuro LC participants showed depletion in multiple genera within the class Clostridia , including Ruminococcus , Lachnospiraceae, and Fusicatenibacter, alongside enrichment of general within order Bacteroidales , including Bacteroides , Phocaeicola , and Alistipes , relative to Mild Neuro and No-Neuro groups; these differences were maintained through 12 months (Fig.1d, left). In males, Severe and Mild Neuro LC participants showed depletion of Alistipes , Roseburia and Bifidobacterium relative to No-Neuro LC participants at both 6 and 12 months, with additional genus-level differences emerging at 12 months (Fig.1d, right). Ordination analyses linked these taxa shifts to specific neurological symptoms, including trouble with memory and concentration, confusion and brain fog, with stronger symptom-taxa associations in females (Fig.1e). Given the higher burden of neurological symptoms among female participants, we tested whether the associated microbiota features were functionally transferable. Germ-free female mice were colonised with stool from female LC donors with neurological symptoms (LC Neuro; n=3 donors) or from female LC donors without neurological symptoms (LC No Neuro; n=3 donors) (Fig.1f). Mice colonized with LC Neuro microbiota showed evidence of impaired intestinal barrier integrity, including reduced/disrupted zonula occludens 1 (ZO-1) staining in colonic tissue (Extended Data Fig.4b), and displayed neurobehavioral differences, including increased locomotion activity in the open-field test (Fig.1g, Extended Data Fig.4a). These mice also developed neuroinflammatory changes, including accumulation of activated astrocytes in the hippocampal region and amoeboid-shaped activated microglia in the hindbrain (Fig.1h,i). In contrast, mice colonized with LC No-Neuro donor microbiota did not show these pathological features. Together, these data indicate that intestinal microbiota associated with neurological symptoms in female LC participants is sufficient to induce intestinal barrier dysfunction, neurobehavioral alteration and neuroinflammatory phenotypes in gnotobiotic mice. GMEVs from individuals with LC induce intestinal epithelial inflammation, impair barrier function and activate macrophages in vitro To test whether GMEVs contribute to LC-associated mucosal and systemic inflammation, we isolated GMEVs from stool pf LC participants with neurological symptoms (LC Neuro), LC participants without neurological symptoms (LC No-Neuro), and pandemic controls (PC). Transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA) confirmed vesicular morphology and broadly similar size distribution across groups, with comparable EV yield metrics as summarized in Extended Data Fig.5. We first assessed epithelial responses to GMEVs using induced pluripotent stem cell–derived human intestinal organoid (HIO) monolayers from healthy donors. Exposure to GMEVs (1 μg/mL) increased inflammasome-associated transcripts in HIOs, including NLRP3 and IL1B , relative to vehicle and PC-derived GMEVs, and also increased TNFSF13B (encoding B-cell activating factor (BAFF)) (Fig.2a). At the protein level, LC-derived GMEVs promoted secretion of inflammatory mediators, with the most pronounced increases generally observed following exposure to LC Neuro GMEVs across measured cytokines and chemokines (including IL-1β, IL-8, TNF, CCL2 and CXCL10) (Fig.2a). Increased intestinal permeability and microbial translocation are hallmarks of chronic intestinal inflammation 22,24 , and individuals with LC exhibit elevated circulating markers of barrier dysfunction and microbial translocation, including soluble zonulin, lipopolysaccharide-binding protein (LBP) and β-D-glucan 34,43 . We therefore evaluated whether GMEVs directly alter epithelial barrier integrity by measuring trans-epithelial electrical resistance (TEER) across Caco-2 cell monolayers. LC Neuro GMEVs produced the strongest reduction in TEER (post-pre), consistent with impaired barrier function, whereas LC No-Neuro GMEVs exerted more modest effects and PC-derived GMEVs had minimal impact (Fig.2b). We next asked whether GMEVs promote innate immune activation that could plausibly contribute to systemic inflammation. THP-1-derived macrophages responded to GMEVs from all donor groups compared to vehicle, consistent with a conserved immunostimulatory capacity of microbiota-derived vesicles (Fig.2c). However, LC Neuro GMEVs elicited the most robust macrophage activation, with higher frequencies and/or intensities of IL-1β-, TNF- and BAFF-associated responses compared with LC No-Neuro and PC conditions (Fig.2c; Extended Data Fig.6d). In parallel, LC-derived GMEVs increased expression of inflammatory transcripts in macrophages, including NLRP3 , TNF and TNFSF13B , with the strongest responses again observed in the LC Neuro condition (Fig.2c). Finally, to compare inflammatory network structure across donor groups, we examined cytokine/chemokine relationships induced by GMEVs. Correlation analyses revealed a denser pattern of coordinated mediator responses in macrophages stimulated with LC Neuro GMEVs than with LC No-Neuro or PC GMEVs, including stronger coupling between TNFSF13B and multiple inflammatory readouts (Extended Data Fig.6a–c). Together, these data indicate that LC-derived GMEVs are sufficient to trigger epithelial inflammatory programmes and macrophage activation in vitro, and that GMEVs from LC Neuro donors most consistently associate with barrier-disruptive activity and coordinated multi-cytokine inflammatory responses. Extracellular vesicles derived from the gut microbiota of individuals with LC induce a pro-inflammatory response in iPSC-derived microglia Because transplantation of LC-associated microbiota induced neuroinflammatory phenotypes in germ-free mice and LC-derived GMEVs activated macrophages in vitro, we asked whether GMEVs can also directly modulate central nervous system innate immune cells. We therefore exposed induced pluripotent stem cell–derived microglia (iMGLs) to donor-derived GMEVs from LC participants (LC Neuro and LC No-Neuro) or pandemic controls (PC) and profiled transcriptional responses by RNA sequencing. Global ordination separated iMGLs stimulated with LC-derived GMEVs from those treated with PC-derived GMEVs, with LC Neuro and LC No-Neuro conditions clustering together, indicating broadly similar microglial responses across LC donor subgroups (Fig.3a). Differential expression analysis identified 2,631 genes altered in iMGLs exposed to LC-derived versus PC-derived GMEVs (adjusted P < 0.05) (Fig.3b). Gene ontology enrichment highlighted a coordinated activation of immune and inflammatory processes, including innate immune response programmes, chemotaxis-related terms and cytokine-linked pathways (Fig.3c; Extended Data Fig.7a). Consistent with this, pathway-level analyses identified significant enrichment of immune and infection -associated signatures, ranked among the top differentially regulated pathways, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) “COVID-19” pathway (Fig.3d; Extended Data Fig.7b,c). Gene set enrichment analysis further showed preferential induction of inflammatory response modules in LC-GMEV-treated iMGLs, including interferon- and cytokine-associated signalling and neutrophil-linked programmes (Fig.3e). In addition, enrichment analyses identified pathways with established roles in neuroimmune communication, including axon guidance, which showed broad upregulation of constituent genes in iMGLs treated with LC-derived GMEVs (Fig.3f). Protein–protein interaction network analysis of highly induced genes revealed interconnected modules centered on inflammatory and myeloid activation nodes, including networks annotated for regulation of neuroinflammatory responses (Extended Data Fig.8a). We validated key transcriptomic changes by targeted qPCR, confirming induction of inflammatory and myeloid-associated genes (for example, IL1B , TNF , IL6 , AIF1 and neutrophil-linked S100 family transcripts) in iMGLs exposed to LC-derived GMEVs compared with PC-derived GMEVs and vehicle (Fig.3g; Extended Data Fig.8b). At the protein level, LC-derived GMEVs increased secretion of inflammatory mediators, including IL-6, TNF and CXCL1 (Fig.3h). Notably, TNFSF13B induction and increased BAFF protein were also observed, paralleling epithelial and macrophage responses and identifying BAFF as a convergent inflammatory mediator in response to GMEVs across multiple cellular compartments (Fig.3g,h). Together, these data show that GMEVs from individuals with LC, irrespective of neurological symptom subgroup, are sufficient to elicit a robust human microglial activation state marked by coordinated induction of inflammatory and neuroimmune-associated gene programmes, thereby supporting a role for LC-associated GMEVs in shaping microglial responses that may contribute to neuroinflammation in LC. Oral administration of GMEVs from individuals with LC remodels the intestinal microbiota in wild-type mice We next asked whether GMEVs are sufficient to reshape intestinal microbial communities in vivo. Wild-type (WT) mice received oral gavage of donor-derived GMEVs (5μg/mL) from LC Neuro participants, LC No-Neuro participants or PCs, or vehicle alone, 3 times per week for 6 weeks. To assess whether orally administered vesicles disseminate beyond the gut, we tracked fluorescently labelled bacterial EVs (derived from Escherichia coli as a surrogate for GMEVs). Eight hours after gavage, fluorescent signal was detected in the gastrointestinal tract and in peripheral organs, including liver and kidney, and was also measurable in brain tissue, whereas signal was not detected in mice receiving unlabelled vesicles (Extended Data Fig.9). To determine whether GMEVs alter intestinal community structure, we performed 16S rRNA gene sequencing on faecal samples collected after 6 weeks of treatment. Relative to vehicle and PC groups, administration of GMEVs from LC donors, irrespective of neurological symptom status, was associated with reduced alpha diversity, with decreases in Shannon diversity and Pielou’s evenness indices (Fig.4a). At the phylum level, mice receiving GMEVs from LC Neuro donors showed the most pronounced compositional shifts, including altered relative abundance of Bacillota to Bacteroidota and reductions in Actinomycetota and Verrucomicrobiota compared to controls groups (Fig.4b). Principal coordinate analysis based on Bray-Curtis dissimilarity showed clear separation of microbiota profiles in mice treated with LC-derived GMEVs from those receiving vehicle or PC-derived GMEVs (Fig.4c). Notably, mice administered GMEVs from LC No-Neuro donors displayed a bifurcated clustering pattern, with partial overlap with both control and LC Neuro clusters, consistent with heterogenous responses to LC No-Neuro GMEVs. We next identified taxa contributing to these treatment-associated differences. Linear discriminant analysis effect size (LEfSe) revealed distinct microbial features associated with each treatment group (Fig.4d). Mice receiving LC Neuro-derived GMEVs were enriched for taxa within the family Lachinospiraceae and the class Clostridia , including the genera Acetatifactor, Lachnobacterium, Fusimonas, Enterocloster, and Velocimicrobium (Fig.4e). In contrast, treatment with LC No-Neuro-derived GMEVs was associated with enrichment of a partially distinct set of genera, including Turicibacter and Bifidobacterium (Fig.4e). Together, these data indicate that oral exposure to LC-derived GMEVs is sufficient to remodel the faecal microbiota in WT mice, with LC Neuro-derived GMEVs associated with the most distinct compositional shifts. Oral administration of GMEVs from individuals with LC induces intestinal inflammation, neurobehavioral alterations and neuroinflammation in vivo We next tested whether oral exposure to donor-derived GMEVs is sufficient to elicit intestinal and systemic inflammatory changes accompanied by neurobehavioural and neuroinflammatory phenotypes in wild-type mice. Female C57BL/6 mice received oral gavage of GMEVs (5 µg per dose) isolated from LC donors with neurological symptoms (Neuro), LC donors without neurological symptoms (No-Neuro) or pandemic controls (PC), or vehicle alone, 3 times per week for 6 weeks, followed by neurobehavioural testing and tissue collection (Fig.5a). Mice receiving LC-derived GMEVs showed evidence of intestinal and systemic inflammatory activation. Colon length was modestly reduced in GMEV-treated groups compared with vehicle, with the largest reduction observed in the Neuro GMEV group (Fig.5b). Circulating markers were also altered as serum lipopolysaccharide-binding protein (LBP) was increased in GMEV-treated groups relative to vehicle, and serum BAFF was significantly elevated in mice receiving Neuro-derived GMEVs compared with controls (Fig.5b). In colonic epithelium, Neuro-derived GMEVs increased expression of inflammasome- and inflammatory-associated genes, including Nlrp3 , Il1b and Tnfsf13b , relative to controls, whereas changes in Tnf and Il6 were more modest (Fig.5c). Consistent with barrier perturbation, ZO-1 immunostaining showed more frequent junctional discontinuities in mice receiving LC-derived GMEVs, which was most apparent in the Neuro GMEV group, relative to vehicle and PC conditions (Fig.5d). We next evaluated whether these inflammatory changes were accompanied by alterations in behaviour. Anxiety-like behaviour was evaluated using the open-field test, and short-term spatial memory was assessed using the Y-maze 44 . In the open-field test, mice receiving Neuro-derived GMEVs travelled a greater distance and displayed higher mean velocity than control groups, indicating increased locomotor activity (Fig.5e). Centre-border measures showed more modest differences as time spent in the border zone increased in the Neuro group, whereas centre time and the centre-to-border time ratio showed non-significant trends (Fig.5e), a pattern that can be suggestive of anxiety-like behaviour but is not definitive in isolation. In the Y-maze, Neuro-derived GMEVs increased total alternations but reduced the proportion of successful alternations, with a concurrent increase in indirect revisits, consistent with altered spontaneous alternation behaviour (Fig.5f) and suggestive of impaired short-term spatial working memory. Memory impairment is a common symptom reported by LC Neuro participants in the IPCO cohort (Fig.1b). Finally, building on the behavioural phenotypes and the observation that orally administered bacterial EVs can be detected beyond the gut (Extended Data Fig.9), we assessed glial activation in brain tissue collected after behavioural testing. Mice treated with Neuro-derived GMEVs showed increased GFAP immunoreactivity, with prominent signal in regions proximal to the hippocampus, and increased IBA1-positive microglial signal with an activated morphology in the hindbrain, compared with control groups (Fig.5g-i). These findings are consistent with induction of neuroinflammatory changes following oral exposure to LC-derived vesicles. Together, these data show that oral administration of GMEVs derived from individuals with LC, most consistently for Neuro-derived preparations, is sufficient to induce intestinal inflammatory responses and barrier perturbation, accompanied by systemic inflammatory signatures, neurobehavioural alterations, and glial activation in vivo. These findings implicate GMEVs as functional mediators linking LC-associated dysbiosis to gut-brain axis perturbations that may contribute to neurological sequelae in LC. Discussion LC is associated with persistent immune dysregulation, intestinal barrier dysfunction, and neuroinflammation 5,8-10,17,20,25 , yet the mechanisms linking these processes remain incompletely defined. Here we identify GMEVs as functional mediators connecting intestinal dysbiosis to systemic and neuroinflammatory features of LC. Using complementary human, in vitro, and in vivo approaches, we show that GMEVs derived from individuals with LC induce epithelial barrier disruption, immune activation, neurobehavioral alterations, and glial activation, thereby providing mechanistic insight into gut-brain axis perturbations in LC. Consistent with previous reports in acute and post-acute syndromes 17-20 , we observed sustained intestinal dysbiosis in individuals with LC that persisted for at least 12 months after infection. Importantly, microbiome alterations correlated with prototypical LC symptoms, particularly neurological manifestations, reinforcing a link between gut microbial composition and disease phenotype. Transfer of LC-associated microbiota into germ-free mice recapitulated features of intestinal barrier dysfunction, systemic inflammation, and neuroinflammation, with effects most pronounced in mice colonization with microbiota from individuals with neurological symptoms. These findings extend prior faecal microbiota transplantation studies 45 by demonstrating symptom-severity-dependent effects and support a contributory role for the gut microbiota in LC pathophysiology. EVs represent a major mode of host-microbe communication, yet their role in post-viral syndromes has remained largely unexplored. Althoug EVs derived from host cells have been implicated in acute COVID-19 and LC 46,47 , our data show that microbiota-derived EVs alone are sufficient to induce LC-relevant phenotypes. GMEVs from LC donors promoted inflammasome activation, epithelial permeability, and inflammatory cytokine/chemokine production in intestinal epithelial cells and macrophages, and induced a coordinated inflammatory transcriptional program in human in iPSC-derived microglia. Notably, while GMEVs from LC Neuro and LC No-Neuro donors elicited broadly similar microglial transcriptomic responses, Neuro-derived GMEVs consistently induced stronger epithelial and macrophage activation, suggesting cell-type-specific sensitivity to GMEV-associated inflammatory signals. In vivo, chronic oral administration of LC-derived GMEVs was sufficient to alter gut microbial composition, disrupt intestinal barrier integrity and induce systemic inflammation, recapitulating several features observed in our microbiota transfer model. Fluorescent tracing experiments demonstrated that orally administered bacterial EVs disseminate beyond the gastrointestinal tract and accumulate in peripheral organs, including the brain, consistent with previous reports 48 . Mice receiving GMEVs from LC Neuro donors developed increased locomotor activity, altered exploratory behaviour and impaired spontaneous alternation performance, accompanied by accumulation of GFAP-positive reactive astrocytes and IBA-1positive activated microglia in discrete brain regions. These findings indicate that LC-derived GMEVs induce spatially localized glial activation consistent with neuroinflammatory processes that may not be captured by bulk tissue analyses. Across several experiments, BAFF emerged as a convergent inflammatory mediator. Elevated BAFF levels distinguished LC severity in our cohort and were consistently induced by LC-derived GMEVs in intestinal epithelial cells, macrophages, microglia and in vivo. Excess BAFF is a recognized driver of chronic immune activation, B-cell dysregulation and autoantibody production 49-54 , and has been implicated in other chronic inflammatory conditions, including HIV infection 49,50,55 . Our findings position BAFF at the intersection of intestinal permeability, immune activation and neuroinflammation in LC, and suggest that GMEVs may contribute to BAFF-driven immunopathology. Mechanistically, GMEVs may engage host cells through multiple pathways, including pattern-recognition receptor signalling triggered by surface-associated microbial components, vesicle internalization, and delivery of biologically active cargo. Differences in EV composition between donor groups, potentially reflecting microbial taxonomic shifts or structural variation in microbial products such as lipopolysaccharide 56,57 , may underlie the graded inflammatory responses observed across GMEVs derived from different LC donor groups. Although LC is undoubtedly multifactorial, our data support a model in which GMEVs amplify and perpetuate inflammatory signalling downstream of intestinal dysbiosis. Several limitations should be considered. Although the IPCO cohort is longitudinal and clinically characterized, LC is heterogeneous and residual confounding by factors that influence the microbiome, including diet, medication exposures and co-morbidities, cannot be fully excluded. The biodistribution experiment used fluorescently-labelled Escherichia coli EVs as a surrogate for complex GMEVs. Thus, the kinetics and tissue tropism of human donor-derived vesicles may differ. Neurobehavioural assays provide operational measures of exploration and spontaneous alternation but do not uniquely map to specific neuropsychiatric constructs. Accordingly, the observed open-field and Y-maze changes should be interpreted as behavioural alterations consistent with neuroimmune perturbation rather than definitive measures of anxiety or memory impairment. Finally, the precise microbial sources and vesicle cargo responsible for host activation were not resolved. Defining the molecular determinants of GMEV bioactivity and the host pathways required for their effects will be important for therapeutic translation. Together, these findings identify GMEVs as previously unrecognized effectors linking intestinal dysbiosis to immune dysregulation and neuroinflammation in LC. By integrating human cohort analyses with mechanistic in vitro and in vivo models, this work highlights GMEVs as both biomarkers and potential therapeutic targets. Interventions aimed at restoring microbial homeostasis or modulating downstream inflammatory pathways, including BAFF signalling, may therefore hold promise for mitigating gut–brain axis dysfunction and neurological symptom burden in long COVID. Methods Study design and population In response to the COVID-19 pandemic, we established the Institut de Recherches Cliniques de Montréal (IRCM) Post-COVID-19 (IPCO) research clinic, which integrates clinical care into a prospective observational cohort study with an associated biobank (IPCO protocol #2021-1092, ClinicalTrials.gov: NCT04736732). Participants are followed longitudinally for up to 24 months with standardized clinical assessments and biospecimen collection. Adults (³18 years) with confirmed SARS-CoV-2 infection at least 3 months before enrolment and persistent symptoms not attributable to alternative diagnoses were recruited as Long COVID (LC) participants. Pandemic controls (PC) were individuals without persistent symptoms and without reported SARS-CoV-2 infection or positive testing. Participants underwent in-person visits at enrolment and at 6, 12 and 24 months after infection, with a telephone follow-up at 18 months. Baseline demographic and clinical characteristics of participants included in microbiome analyses are provided in Extended Data Table 1. Symptom definitions and neurological stratification Participants reported symptoms using standardized questionnaires administered at each visit. For analyses stratifying LC participants by neurological symptom burden, 8 self-reported symptoms were considered: post-exertional malaise, trouble with concentration, trouble with memory, trouble with sleep, anxiety, brain fog, confusion and depression. Participants reporting ³1 symptom were classified as LC Neuro, whereas those reporting non were classified LC No-Neuro. For symptom-burden analyses, LC participants reporting symptoms were categorized as Mild Neuro (<3 symptoms) or Severe Neuro (≥3 symptoms). Sample collection and processing At each study visit, participants provided stool, blood, saliva, and urine samples. Serum and plasma were isolated by centrifugation. Peripheral blood mononuclear cells (PBMCs) were isolated using SepMate tubes (STEMCELL Technologies). All biospecimens were aliquoted and stored at -80°C, and PBMCs were cryopreserved in liquid nitrogen. Stool DNA extraction and microbiome sequencing Human samples Stool DNA was extracted using the QIAamp PowerFecal Pro DNA Kit (QIAGEN) with mechanical homogenization. Shotgun metagenomic libraries were prepared after quality control and sequenced on an Illumina NovaSeq 6000 platform (CosmosID). Host reads were removed by alignment to the human reference genome (GRCh38). Read quality was assessed using FastQC 58 . Functional profiling was performed using HUMAnM (v3.9) 59 and taxonomic profiling using MetaPhlAn (v4.1.1) 60 . Microbiome analyses were performed in R (v4.2) using phyloseq (v1.48.0) 61 and tidyverse packages 62 . Alpha diversity was assessed using the Simpson index Pielou’s evenness, and beta diversity using Bray-Curtis dissimilarity with principal coordinate analysis (PCoA). Differentially abundant taxa were identified using Linear Discriminant Analysis (LDA) Effect Size (LEfSe)-based approaches implemented in edgeR 63 and microbiomeMarker 64 , with a Kruskal-Wallis significance threshold of P ≤ 0.05 and an LDA score cutoff as indicated in the corresponding figure legends. Mouse stool 16S rRNA sequencing Mouse faecal pellets were collected into sterile tubes, snap-frozen on dry ice and stored at -80°C. DNA was extracted using the DNeasy PowerSoil Pro QIAcube HT Kit (QIAGEN). The V4 region of the 16S rRNA gene was amplified and sequenced on an Illumina MiSeq platform (McGill Centre for Microbiome Research). Read quality was assessed using FastQC 58 , adapters and primers were trimmed with Trimmomatic 65 . Paired-end reads were merged with PANDAseq 66 and chimeras removed as described previously 67 . Sequences were clustered using CD-HIT-EST 68 and taxonomically classified using Kraken2 69 against an in-house RefSeq database (updated November 2023). Alpha diversity was assessed using Shannon diversity and Pielou’s evenness. Beta diversity was evaluated using Bray-Curtis dissimilarity and visualized by PCoA. Group differences were assessed by permutational multivariate analysis of variance (PERMANOVA; adonis2, vegan v2.6-6.1; 1000 permutations) 70 , with pairwise comparisons adjusted for multiple testing as indicated. Differentially abundant taxa were identified using LEfSe (edgeR 63 and microbiomeMarker 64 ; LDA cutoff and significance thresholds as indicated in the corresponding figure legends). Animals and housing Wild-type (WT) C57BL/6 female mice (6-8 weeks old) were obtained from The Jackson Laboratory and housed under specific pathogen-free conditions at the IRCM animal facility with a 12-hour light/dark cycle with controlled temperature and humidity and ad libitum access to food and water. Germ-free (GF) C57BL/6 female mice (5-7 weeks old) were obtained from the Germ-Free and Gnotobiotic Platform at the University of Calgary (K. McCoy) and maintained in flexible film isolators (Class Biologically Clean) under sterile conditions at the IRCM gnotobiotic facility. Mice were routinely screened for contamination. All animal procedures were approved by the IRCM Animal Care Committee. Faecal microbiota transplantation into germ-free mice Mouse experimental group sizes and anonymised donor sample identifiers are summarized in Extended Data Table 2. GF mice were orally gavaged 3 times per week with 200μl of a stool suspension prepared by resuspending 200mg of human donor stool in 1mL sterile PBS. After the first gavage, mice were transferred to positive-pressure cages to prevent cross-contamination. Microbiota were allowed to engraft for 3 weeks before mice were transferred to irradiated racks for acclimation, behavioural testing and tissue collection. GMEV isolation and characterization Gut microbiota-derived extracellular vesicles (GMEVs) were isolated from human stool by size-exclusion chromatography. Stool suspensions were sequentially centrifuged and filtered (0.45μm and 0.22μm) to remove bacteria and debris, concentrated by ultrafiltration (100kDa cut-off), and fractionated using qEV Original 35nm Gen 2 column (IZON). GMEV-enriched fractions were pooled, aliquoted, and stored at -80°C. Particle size and concentration were assessed by nanoparticle tracking analysis (ZetaView PMX120). Vesicular morphology was confirmed by transmission electron microscopy (Tecnai G2 Spirit Twin). Protein concentration was measured using the Pierce Bovine Serum Albumin Protein assay (Thermo Fisher Scientific). Neurobehavioural testing Neurobehavioural testing was performed after microbiota engraftment (GF colonization experiments) or after the oral gavage regimen (WT GMEV experiments), as indicated in the relevant figure legends. Testing was recorded using the EthoVision and analysed blinded to group allocation where feasible. Y-maze spontaneous alternation The Y-maze test was used to assess spontaneous alternation behaviour, which is sensitive to short-term spatial working memory 44 . Mice were placed in the centre of a 3-armed maze and allowed to explore freely for 8 minutes. Arm entries were scored by EthoVision. Consecutive triplets of arm entries were defined as alternations; triplets comprising 3 different arms were scored as successful alternations. Performance was calculated as the ratio of successful alternations to total alternations, with direct and indirect revisits also quantified. A lux meter was used to ensure that illumination was standardized across all 3 arms (approximately 150 lux). Open-field test Exploration and anxiety-like behaviour were assessed using the open-field test 71 . Mice were placed in the arena and recorded by EthoVision. Time spent in the centre versus periphery (border), the number of centre-border transitions, total distance travelled, and velocity were quantified. A lux meter was used to ensure that illumination was standardized across the whole arena (approximately 150 lux). Oral gavage of GMEVs in wild-type mice Group sizes and anonymised donor sample identifiers are provided in Extended Data Table 2. Female WT C56BL/6 mice (6-8 weeks old; The Jackson Laboratory), maintained under specific pathogen-free conditions, received GMEVs by oral gavage (5μg per dose in 200μl sterile PBS) or vehicle (sterile PBS) 3 times per week for 6 weeks. Following the treatment period, mice underwent behavioural testing prior to euthanasia and tissue collection. Colon length was measured at necropsy. Colon and brain tissue were processed for histology and molecular analyses as described below. Biodistribution of orally administered EVs To assess biodistribution following oral delivery, extracellular vesicles were isolated from Escherichia coli (ATCC 25922) culture supernatants by ultrafiltration and size-exclusion chromatography (qEV 35 nm Gen 2; Izon). Vesicles were labelled with Vybrant DiD (Invitrogen) according to the manufacturer’s instructions and administered to female BALB/c mice by oral gavage (10µg dose of EVs). Fluorescence was acquired 8h after gavage using a Xenogen IVIS 200 system (PerkinElmer) under isoflurane anaesthesia. Mice were then euthanized, and organs were harvested for ex vivo fluorescence imaging. Blood collection and serum isolation Mice were euthanized by terminal cardiac puncture. Whole blood was allowed to clot at room temperature for 30 minutes, incubated at 37°C for 10 minutes, and then incubated at 4°C for 15 minutes before centrifugation (3,000g, 20 minutes). Serum was stored at -80°C until analysis. ELISA measurements of soluble BAFF and LBP Serum B-cell activation factor (BAFF) and lipopolysaccharide-binding protein (LBP) were quantified using the Quantikine Mouse BAFF/BLyS/TNFSF13B ELISA kit (R&D Systems) and the Mouse LBP ELISA kit (Abcam), respectively, following the manufacturers’ instructions. Colon epithelial cell isolation Colon epithelial cells were isolated using a dithiothreitol (DTT)/EDTA-based epithelial stripping approach. Colons were washed in HBSS without Ca ++ and Mg ++ supplemented with EDTA (2mM) and HEPES (25mM). Tissue was incubated at 37°C for 15 minutes in stripping buffer (HBSS without Ca ++ /Mg ++ supplemented with HEPES (15mM), EDTA (5mM) and DTT (1mM)) and vortexed to release epithelial cells. The epithelial fraction was filtered (100μm), washed in PBS, and lysed in RLT buffer (QIAGEN) for RNA extraction. Tissue embedding, immunofluorescence staining and imaging Colon tissue and brain hemisphere were embedded in OCT embedding medium (Scigen Scientific) and stored at -80°C. Colon cryosections (10µm) were prepared at the IRCM Histology Core Facility, fixed in cold acetone (-20°C), air dried, and stored at -80°C. Cryosections (14μm) of brain hemispheres were prepared at the Institut de Recherche en Immunologie et en Cancérologie (IRIC) Histology Core Facility and processed identically. Immunofluorescence staining was performed at the Centre de Recherche du Centre Hospitalier de l’Université de Montréal (CRCHUM) Molecular Pathology Core Facility using a Discovery Ultra automated stainer (Ventana/Roche). Sections were blocked in PBS containing 1% bovine serum albumin (BSA) for 30 minutes at room temperature, incubated with primary antibodies followed by species-appropriate secondary antibodies for 2 hours each at room temperature. Nuclei were counterstained with DAPI (1:3,000) for 10 minutes. Sudan Black (0.1% in 70% ethanol) was applied to reduce autofluorescence, and sections were mounted with Fluoromount (Sigma). Slides were scanned using an Aperio Verso 200 scanner (Leica Biosystems) with a 20´/0.8 NA objective at a resolution of 0.275μm per pixel. Image visualization and analysis were performed blinded to group allocation using Aperio ImageScope software (Leica Biosystems). Primary antibodies used included rabbit anti-mouse zonula occludens-1 (ZO-1) and ionized calcium-binding adaptor molecule 1 (IBA1) and an anti-glial fibrillary acidic protein (GFAP) monoclonal antibody; Alexa Fluor-conjugated goat anti-rabbit secondary antibodies were used as indicated. Antibody details are provided in Extended Data Table 4b. Cell culture and stimulation with GMEVs THP-1-derived human macrophages THP-1 cells were maintained in complete RPMI 1640 supplemented with 10% heat-inactivated foetal bovine serum (FBS), 1% penicillin-streptomycin and 14.3μM 2-mercaptoethanol. For differentiation, THP-1 cells were seeded at 5 x 10 5 cells per well (12-well plates) and treated with phorbol 12-myristate 13-acetate (PMA; 100nM) for 48 hours, washed and rested in fresh complete medium before stimulation. Macrophages were stimulated with GMEVs (1μg/mL) for 5h (RNA) or 16h (secreted proteins). For intracellular cytokine staining, cells were treated with BD GolgiPlug protein transport inhibitor (containing brefeldin A) during the final 5h of stimulation. Human intestinal organoids and epithelial monolayers Human intestinal organoids (HIOs) were generated from human induced pluripotent stem cells (iPSCs) derived from fibroblasts from a healthy female donor (generously provided by H. Malech, NIH) using the STEMdiff™ Intestinal Organoid Kit (STEMCELL Technologies) following the manufacturer’s instructions. iPSCs were maintained on Matrigel in mTeSR1 medium and differentiated into intestinal lineage using the STEMdiff™ Intestinal Organoid Kit following the manufacturer’s protocol, including definitive endoderm induction and subsequent mid-hindgut patterning. Mid-hindgut spheroids were embedded in Matrigel domes and matured in Intestinal Organoid Growth Medium, with medium changes every 2-3 days and weekly passaging. For stimulation experiments, HIOs were dissociated into epithelial monolayers, plated onto Matrigel-coated 24-well plates and treated with GMEVs (1 µg/mL) for 5h (RNA) or 16h (secreted proteins). Transepithelial electrical resistance Caco-2 cells were seeded at 1.5x10 5 cells/cm 2 on Transwell inserts (0.4μm pore; 12-well format; Sarstedt) and maintained for ≥3 weeks until transepithelial electrical resistance (TEER) stabilized. Cells were cultured in Eagle’s Modified Essential Medium (EMEM) supplemented with 20% heat-inactivated FBS and penicillin-streptomycin. Baseline TEER was recorded before treatment. Cells were treated with GMEVs (3 µg/mL), and TEER measured 24h later using an EVOM2 voltohmmeter with STX2 electrodes. Values reflect the mean of 3 measurements per insert. RNA Extraction and quantitative PCR Total RNA was extracted from colon epithelial cells, HIOs, iPSC-derived microglia, and THP-1-derived macrophages using the RNeasy Plus Mini Kit (QIAGEN) according to the manufacturer's instructions. RNA concentration and purity were assessed using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). Complementary DNA was synthesized from 1μg total RNA using the iScript Reverse Transcription SuperMix for RT-qPCR (BioRad). Quantitative PCR was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad) on a StepOnePlus Real-Time PCR System (Applied Biosystems). Relative gene expression was calculated using the 2 (-ΔΔCt) method with GAPDH as the reference gene. Primer sequences are listed in Extended Data Table 3. Differentiation of induced pluripotent stem cell-derived microglia Human iPSCs were generated from a healthy female donor and maintained using standard procedures as described 72 . Microglia were differentiated from iPSCs using a two-step protocol adapted from established methods 73 . Briefly, iPSCs were differentiated into iPSC-derived hematopoietic progenitor cells (iHPCs) using the STEMdiff Hematopoietic Kit (STEMCELL Technologies), with minor modifications. On day -1, iPSCs were dissociated using Gentle Cell Dissociation Reagent (STEMCELL Technologies) and seeded onto Matrigel-coated 6-well plates in mTeSR™ Plus or Essential 8 medium supplemented with Y-27632 (10μM, Selleckchem) at a density designed to yield colonies of fewer than 100 cells per cm² by day 0. On day 0, medium (1mL) was replaced with STEMdiff Hematopoietic Medium A (2mL per well). On day 2, half of the medium (1mL) was replaced with fresh Medium A. On day 3, cultures were switched entirely to STEMdiff Hematopoietic Medium B (2mL per well). On days 5 and 7, half of the supernatant (1mL) was replaced with fresh Medium B. On day 9, an additional 1mL of fresh Medium B was added. On day 10, non-adherent iHPCs present in the supernatant were collected, centrifuged at 300g for 5 minutes, and either cryopreserved (Bambanker, Fujifilm Wako Chemicals) or used for microglial differentiation. This harvesting procedure was repeated on days 12 and 14. For microglial differentiation, iHPCs were resuspended in microglia differentiation medium (as previously described) at a density of 5x10 4 cells/mL and plated onto Matrigel-coated 6-well plates (2mL per well), defining day 0 of differentiation. Cultures were supplemented every other day with 1mL of fresh differentiation medium from day 0 to day 10. On day 12, supernatants were collected, cells were pelleted by centrifugation (300g, 5 minutes), resuspended in fresh differentiation medium, and returned to culture. This procedure was repeated on day 24. Cells were considered mature by day 28 and were maintained with media changes every other day until use. All cultures were maintained at 37°C in a humidified atmosphere containing 5% CO 2 . For downstream experiments, cells were detached using PBS containing 2mM EDTA and replated at the desired density. Bulk RNA sequencing of GMEV-treated iMGLs Induced pluripotent stem cell-derived microglia (iMGLs) were treated with GMEVs (1μg/mL) from LC donors (n=10) or PC donors (n=4) for 16h. RNA (£100ng) was submitted for ribodepletion library preparation and sequencing (target ~50 million reads per sample; IRCM Molecular Biology platform). Read quality was assessed using FastQC (v0.12.1). Reads were aligned to GRCh38 using STAR (v2.7.11b), gene counts were generated using featureCounts v2.0.6; GRCh38 release 110), and differential expression was performed using the DESeq2. Heatmaps were generated using z-scored normalized counts. Functional enrichment analysis was performed using gprofiler2. Cytokine and chemokine quantification in cell culture supernatants Secreted cytokines/chemokines were quantified using MSD U-PLEX panels. For THP-1 macrophages and HIOs, analytes included granulocyte-macrophage colony stimulating factor (GM-CSF), granulocyte-colony stimulating factor (G-CSF), interferon gamma (IFNγ), IL-10, IL-1β, IL-6, IL-8, CXCL10, CCL2, and tumor necrosis factor (TNF). For iMGL supernatants, the panel included B-cell activation factor (BAFF), IL-1β, IL-6, IL-8, TNF, s100A12, matrix metalloproteinase 9 (MMP9), CCL2 and CXCL1, with R-PLEX detection of C1q. Plates were read on a MESO QuickPlex SQ 120 and analysed using MSD Discovery Workbench 4.0. Flow cytometry of THP-1-derived macrophages Cells were stained with LIVE/DEAD Fixable Aqua (Thermo Fisher Scientific), blocked with human Fc block (BD Biosciences) with 20% heat inactivated FBS and 50μg mouse and/or rat IgG, and stained for surface BAFF, followed by fixation/permeabilization (Cytofix/Cytoperm, BD Biosciences) and intracellular staining for IL-1ꞵ, IL-6 and TNF. Data were acquired on a BD LSRFortessa and analysed in FlowJo (v10.8.1). Antibody details are provided in Extended Data Table 4a. Statistical analyses Group comparisons were performed using one-way ANOVA with Tukey’s post hoc test for approximately normally distributed data, or Kruskal–Wallis tests with Dunn’s post hoc correction otherwise. For two-group comparisons, Wilcoxon rank-sum (two-sided) tests were used unless stated otherwise. Correlations were assessed using Pearson’s or Spearman’s tests, as appropriate. Categorical variables were analysed using Fisher’s exact test. Unless otherwise indicated, data are shown as medians with interquartile ranges. All tests were two-sided and P < 0.05 was considered statistically significant. Analyses were performed using GraphPad Prism v10.2.0. Declarations Acknowledgements E.L.F is supported by a Tier 2 Canada Research Chair in Role of the Microbiome in Inborn Errors of Immunity and Post-Infectious Conditions, the Canadian Institutes of Health Research (CIHR), the Fonds de Recherche du Québec (FRQ) . K.D.L. is supported by the IRCM Foundation. M.A. is supported by CIHR. A.D. is supported by the Fonds de Recherche du Québec (FRQ) . I.B. is supported by the IRCM Foundation. The work was supported by CIHR (PJT-191724), the Canada Research Chairs Program, the FRQ Clinical Research Scholars - Junior 1 Establishment Funds for Young Investigators, the John R. Evans Leaders Fund from the Canadian Foundation for Innovation (CFI), the J-Louis Lévesque Foundation Research Chair, the Mirella and Lino Saputo Foundation, and the IRCM Foundation. We thank members of the IRCM animal facility (Mariane Canuel, Eve-Marie Charbonneau, Manon Laprise, and Jo-Anny Bisson), the IRCM Flow Cytometry Platform (Éric Massicotte and Julie Lord), and the McGill Genome Centre, Centre for Microbiome Research, for technical support. We are grateful to Kathy McCoy for providing germ-free mice. We thank Christian Charbonneau, Dr. Marianne Isaac and Melina Narlis of the IRIC Microscopy and Histology Core Facilities for guidance and for performing sagittal brain cryosections and H&E staining. We also thank Véronique Barrès and Liliane Meunier of the CRCHUM Molecular Pathology Core Facility for immunolabeling, slide scanning, and assistance with paraffin and OCT processing of colon and brain, and Anabelle Bouchard-Bourque of the IRCM Histology Core Facility for additional paraffin and OCT sectioning. We acknowledge Dominic Filion and Mattew Duguay of the IRCM Imaging Platform for imaging support. We thank Dr. Mieczyslaw Marcinkiewicz, and Dennis A. Drewnik for advice and assistance with brain immunofluorescence interpretation, fixation and embedding. We also thank the Centre for Applied Nanomedicine at the Research Institute of the McGill University Hospital Centre for assistance with nanoparticle tracking analysis, and Dr. S. Kelly Sears and Dr. Jeannie Mui from the Facility for Electron Microscopy Research at McGill University for their assistance with electron microscopy. Author contributions M.A. and E.L.F. conceived the study and designed the research. M.A., K.D-L., I.B., A.V., A.D., E.D, L.P., V.E.P., F.B., J.S., and A.A. performed experiments. M.A., K.D-L., I.B., P.C., J.P. and E.L.F. analysed the data. M.A., P.C., and E.L.F. wrote the initial manuscript draft, and all authors contributed to manuscript revision. E.L.F. secured funding for the study. E.L.F. supervised the research. Competing interests The authors declare no competing interest. Data availability All raw sequencing data has been deposited in SRA under BioProject PRJNA1236664 (submissions: SUB15113758 and SUB15164815) and will be released upon publication. Any additional data supporting the findings of this study are available from the corresponding author upon reasonable request. Code availability Custom code used for microbiota processing, statistical analyses, and figure generation will be deposited on GitHub and made publicly available upon publication. Scripts are available from the corresponding author upon reasonable request prior to release. References World Health, O. A clinical case definition of post COVID-19 condition by a Delphi consensus. (World Health Organization, Geneva, 2021). Ely, E. W., Brown, L. M., Fineberg, H. V., National Academies of Sciences, E. & Medicine Committee on Examining the Working Definition for Long, C. Long Covid Defined. N Engl J Med 391 , 1746-1753 (2024). https://doi.org/10.1056/NEJMsb2408466 Taquet, M. et al. Neurological and psychiatric risk trajectories after SARS-CoV-2 infection: an analysis of 2-year retrospective cohort studies including 1 284 437 patients. The Lancet Psychiatry 9 , 815-827 (2022). https://doi.org/10.1016/S2215-0366(22)00260-7 Davis, H. E., McCorkell, L., Vogel, J. M. & Topol, E. J. Long COVID: major findings, mechanisms and recommendations. Nat Rev Microbiol 21 , 133-146 (2023). https://doi.org/10.1038/s41579-022-00846-2 Kuang, S. E., S.; Clarke, J.; Zakaria, D.; Demers, A.; Aziz, S. Experiences of Canadians with long-term symptoms following COVID-19. Insights on Canadian Society (2023). COVID-19, W. H. O. W. c. c. d. w. g. o. p. & condition. (2021). Davis, H. E. et al. Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. EClinicalMedicine 38 , 101019 (2021). https://doi.org/10.1016/j.eclinm.2021.101019 Lopez-Leon, S. et al. More than 50 long-term effects of COVID-19: a systematic review and meta-analysis. Sci Rep 11 , 16144 (2021). https://doi.org/10.1038/s41598-021-95565-8 Al-Aly, Z., Xie, Y. & Bowe, B. High-dimensional characterization of post-acute sequelae of COVID-19. Nature 594 , 259-264 (2021). https://doi.org/10.1038/s41586-021-03553-9 Al-Aly, Z. & Topol, E. Solving the puzzle of Long Covid. Science 383 , 830-832 (2024). https://doi.org/10.1126/science.adl0867 Cai, M., Xie, Y., Topol, E. J. & Al-Aly, Z. Three-year outcomes of post-acute sequelae of COVID-19. Nat Med 30 , 1564-1573 (2024). https://doi.org/10.1038/s41591-024-02987-8 Klein, J. et al. Distinguishing features of long COVID identified through immune profiling. Nature 623 , 139-148 (2023). https://doi.org/10.1038/s41586-023-06651-y Yin, K. et al. Long COVID manifests with T cell dysregulation, inflammation and an uncoordinated adaptive immune response to SARS-CoV-2. Nat Immunol 25 , 218-225 (2024). https://doi.org/10.1038/s41590-023-01724-6 Phetsouphanh, C. et al. Immunological dysfunction persists for 8 months following initial mild-to-moderate SARS-CoV-2 infection. Nat Immunol 23 , 210-216 (2022). https://doi.org/10.1038/s41590-021-01113-x Woodruff, M. C. et al. Chronic inflammation, neutrophil activity, and autoreactivity splits long COVID. Nat Commun 14 , 4201 (2023). https://doi.org/10.1038/s41467-023-40012-7 Talla, A. et al. Persistent serum protein signatures define an inflammatory subcategory of long COVID. Nat Commun 14 , 3417 (2023). https://doi.org/10.1038/s41467-023-38682-4 Pretorius, E. et al. Persistent clotting protein pathology in Long COVID/Post-Acute Sequelae of COVID-19 (PASC) is accompanied by increased levels of antiplasmin. Cardiovasc Diabetol 20 , 172 (2021). https://doi.org/10.1186/s12933-021-01359-7 Braga, J. et al. Neuroinflammation After COVID-19 With Persistent Depressive and Cognitive Symptoms. JAMA Psychiatry 80 , 787-795 (2023). https://doi.org/10.1001/jamapsychiatry.2023.1321 Soung, A. L. et al. COVID-19 induces CNS cytokine expression and loss of hippocampal neurogenesis. Brain 145 , 4193-4201 (2022). https://doi.org/10.1093/brain/awac270 Chang, R. et al. Risk of autoimmune diseases in patients with COVID-19: A retrospective cohort study. EClinicalMedicine 56 , 101783 (2023). https://doi.org/10.1016/j.eclinm.2022.101783 Son, K. et al. Circulating anti-nuclear autoantibodies in COVID-19 survivors predict long COVID symptoms. Eur Respir J 61 (2023). https://doi.org/10.1183/13993003.00970-2022 Nayyerabadi, M. et al. Vaccination after developing long COVID: Impact on clinical presentation, viral persistence, and immune responses. Int J Infect Dis 136 , 136-145 (2023). https://doi.org/10.1016/j.ijid.2023.09.006 Swank, Z. et al. Persistent Circulating Severe Acute Respiratory Syndrome Coronavirus 2 Spike Is Associated With Post-acute Coronavirus Disease 2019 Sequelae. Clin Infect Dis 76 , e487-e490 (2023). https://doi.org/10.1093/cid/ciac722 Natarajan, A. et al. Gastrointestinal symptoms and fecal shedding of SARS-CoV-2 RNA suggest prolonged gastrointestinal infection. Med 3 , 371-387 e379 (2022). https://doi.org/10.1016/j.medj.2022.04.001 Stein, S. R. et al. SARS-CoV-2 infection and persistence in the human body and brain at autopsy. Nature 612 , 758-763 (2022). https://doi.org/10.1038/s41586-022-05542-y Zollner, A. et al. Postacute COVID-19 is Characterized by Gut Viral Antigen Persistence in Inflammatory Bowel Diseases. Gastroenterology 163 , 495-506 e498 (2022). https://doi.org/10.1053/j.gastro.2022.04.037 Yeoh, Y. K. et al. Gut microbiota composition reflects disease severity and dysfunctional immune responses in patients with COVID-19. Gut 70 , 698-706 (2021). https://doi.org/10.1136/gutjnl-2020-323020 Liu, Q. et al. Gut microbiota dynamics in a prospective cohort of patients with post-acute COVID-19 syndrome. Gut 71 , 544-552 (2022). https://doi.org/10.1136/gutjnl-2021-325989 Honda, K. & Littman, D. R. The microbiota in adaptive immune homeostasis and disease. Nature 535 , 75-84 (2016). https://doi.org/10.1038/nature18848 Malard, F., Dore, J., Gaugler, B. & Mohty, M. Introduction to host microbiome symbiosis in health and disease. Mucosal Immunol 14 , 547-554 (2021). https://doi.org/10.1038/s41385-020-00365-4 Di Vincenzo, F., Del Gaudio, A., Petito, V., Lopetuso, L. R. & Scaldaferri, F. Gut microbiota, intestinal permeability, and systemic inflammation: a narrative review. Internal and Emergency Medicine 19 , 275-293 (2024). https://doi.org/10.1007/s11739-023-03374-w Giron, L. B. et al. Plasma Markers of Disrupted Gut Permeability in Severe COVID-19 Patients. Front Immunol 12 , 686240 (2021). https://doi.org/10.3389/fimmu.2021.686240 Prasad, R. et al. Plasma Microbiome in COVID-19 Subjects: An Indicator of Gut Barrier Defects and Dysbiosis. Int J Mol Sci 23 (2022). https://doi.org/10.3390/ijms23169141 Giron, L. B. et al. Markers of fungal translocation are elevated during post-acute sequelae of SARS-CoV-2 and induce NF-kappaB signaling. JCI Insight 7 (2022). https://doi.org/10.1172/jci.insight.160989 Morais, L. H., Schreiber, H. L. t. & Mazmanian, S. K. The gut microbiota-brain axis in behaviour and brain disorders. Nat Rev Microbiol 19 , 241-255 (2021). https://doi.org/10.1038/s41579-020-00460-0 Ancuta, P. et al. Microbial translocation is associated with increased monocyte activation and dementia in AIDS patients. PLoS One 3 , e2516 (2008). https://doi.org/10.1371/journal.pone.0002516 Chen, X., Wei, J., Zhang, Y., Zhang, Y. & Zhang, T. Crosstalk between gut microbiome and neuroinflammation in pathogenesis of HIV-associated neurocognitive disorder. J Neurol Sci 457 , 122889 (2024). https://doi.org/10.1016/j.jns.2024.122889 Leclerc, L., Poudrier, J., Power, C., Lam, G. Y. & Falcone, E. L. Intestinal barrier compromise, viral persistence, and immune dysregulation converge on neurological sequelae in Long COVID. Front Aging Neurosci 17 , 1744415 (2025). https://doi.org/10.3389/fnagi.2025.1744415 Xie, J., Haesebrouck, F., Van Hoecke, L. & Vandenbroucke, R. E. Bacterial extracellular vesicles: an emerging avenue to tackle diseases. Trends Microbiol 31 , 1206-1224 (2023). https://doi.org/10.1016/j.tim.2023.05.010 Villard, A., Boursier, J. & Andriantsitohaina, R. Microbiota-derived extracellular vesicles and metabolic syndrome. Acta Physiol (Oxf) 231 , e13600 (2021). https://doi.org/10.1111/apha.13600 Villard, A., Boursier, J. & Andriantsitohaina, R. Bacterial and eukaryotic extracellular vesicles and nonalcoholic fatty liver disease: new players in the gut-liver axis? Am J Physiol Gastrointest Liver Physiol 320 , G485-G495 (2021). https://doi.org/10.1152/ajpgi.00362.2020 Fizanne, L. et al. Faeces-derived extracellular vesicles participate in the onset of barrier dysfunction leading to liver diseases. J Extracell Vesicles 12 , e12303 (2023). https://doi.org/10.1002/jev2.12303 Fasano, A. Zonulin and its regulation of intestinal barrier function: the biological door to inflammation, autoimmunity, and cancer. Physiol Rev 91 , 151-175 (2011). https://doi.org/10.1152/physrev.00003.2008 Kraeuter, A. K., Guest, P. C. & Sarnyai, Z. The Y-Maze for Assessment of Spatial Working and Reference Memory in Mice. Methods in molecular biology 1916 , 105-111 (2018). Mendes de Almeida, V. et al. Gut microbiota from patients with COVID-19 cause alterations in mice that resemble post-COVID symptoms. Gut Microbes 15 , 2249146 (2023). https://doi.org/10.1080/19490976.2023.2249146 Forte, D. et al. Circulating extracellular particles from severe COVID-19 patients show altered profiling and innate lymphoid cell-modulating ability. Front Immunol 14 , 1085610 (2023). https://doi.org/10.3389/fimmu.2023.1085610 Nair, S. et al. Genomic communication via circulating extracellular vesicles and long-term health consequences of COVID-19. J Transl Med 21 , 709 (2023). https://doi.org/10.1186/s12967-023-04552-2 Jones, E. J. et al. The Uptake, Trafficking, and Biodistribution of Bacteroides thetaiotaomicron Generated Outer Membrane Vesicles. Front Microbiol 11 , 57 (2020). https://doi.org/10.3389/fmicb.2020.00057 Doyon-Laliberte, K. et al. Excess BAFF Alters NR4As Expression Levels and Breg Function of Human Precursor-like Marginal Zone B-Cells in the Context of HIV-1 Infection. Int J Mol Sci 23 (2022). https://doi.org/10.3390/ijms232315142 Aranguren, M. et al. Subclinical Atherosclerosis Is Associated with Discrepancies in BAFF and APRIL Levels and Altered Breg Potential of Precursor-like Marginal Zone B-Cells in Long-Term HIV Treated Individuals. Vaccines (Basel) 11 (2022). https://doi.org/10.3390/vaccines11010081 Becker-Merok, A., Nikolaisen, C. & Nossent, H. C. B-lymphocyte activating factor in systemic lupus erythematosus and rheumatoid arthritis in relation to autoantibody levels, disease measures and time. Lupus 15 , 570-576 (2006). https://doi.org/10.1177/0961203306071871 Chu, V. T. et al. Systemic activation of the immune system induces aberrant BAFF and APRIL expression in B cells in patients with systemic lupus erythematosus. Arthritis & Rheumatism 60 , 2083-2093 (2009). https://doi.org/https://doi.org/10.1002/art.24628 Steri, M. et al. Overexpression of the Cytokine BAFF and Autoimmunity Risk. N Engl J Med 376 , 1615-1626 (2017). https://doi.org/10.1056/NEJMoa1610528 Lesley, R. et al. Reduced competitiveness of autoantigen-engaged B cells due to increased dependence on BAFF. Immunity 20 , 441-453 (2004). https://doi.org/10.1016/s1074-7613(04)00079-2 Doyon-Laliberté, K. et al. Excess BAFF May Impact HIV-1-Specific Antibodies and May Promote Polyclonal Responses Including Those from First-Line Marginal Zone B-Cell Populations. Curr Issues Mol Biol 46 , 25-43 (2023). https://doi.org/10.3390/cimb46010003 Steimle, A., Autenrieth, I. B. & Frick, J.-S. Structure and function: Lipid A modifications in commensals and pathogens. International Journal of Medical Microbiology 306 , 290-301 (2016). https://doi.org/https://doi.org/10.1016/j.ijmm.2016.03.001 Behrouzi, A. et al. Comparative study of pathogenic and non-pathogenic Escherichia coli outer membrane vesicles and prediction of host-interactions with TLR signaling pathways. BMC Res Notes 11 , 539 (2018). https://doi.org/10.1186/s13104-018-3648-3 FastQC: a quality control tool for high throughput sequence data (2010). Franzosa, E. A. et al. Species-level functional profiling of metagenomes and metatranscriptomes. Nat Methods 15 , 962-968 (2018). https://doi.org/10.1038/s41592-018-0176-y Manghi, P. et al. MetaPhlAn 4 profiling of unknown species-level genome bins improves the characterization of diet-associated microbiome changes in mice. Cell Reports 42 , 112464 (2023). https://doi.org/https://doi.org/10.1016/j.celrep.2023.112464 McMurdie, P. J. & Holmes, S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. PLoS One 8 , e61217 (2013). https://doi.org/10.1371/journal.pone.0061217 Whitmore, N. R for Conservation and Development Projects A Primer for Practitioners . (Chapman and Hall / CRC Press, 2022). Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26 , 139-140 (2010). https://doi.org/10.1093/bioinformatics/btp616 Cao, Y. et al. microbiomeMarker: an R/Bioconductor package for microbiome marker identification and visualization. Bioinformatics 38 , 4027-4029 (2022). https://doi.org/10.1093/bioinformatics/btac438 Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30 , 2114-2120 (2014). https://doi.org/10.1093/bioinformatics/btu170 Masella, A. P., Bartram, A. K., Truszkowski, J. M., Brown, D. G. & Neufeld, J. D. PANDAseq: paired-end assembler for illumina sequences. BMC Bioinformatics 13 , 31 (2012). https://doi.org/10.1186/1471-2105-13-31 Mlaga, K. D. et al. HCK and ABAA: A Newly Designed Pipeline to Improve Fungi Metabarcoding Analysis. Front Microbiol 12 , 640693 (2021). https://doi.org/10.3389/fmicb.2021.640693 Fu, L., Niu, B., Zhu, Z., Wu, S. & Li, W. CD-HIT: accelerated for clustering the next-generation sequencing data. Bioinformatics 28 , 3150-3152 (2012). https://doi.org/10.1093/bioinformatics/bts565 Lu, J. & Salzberg, S. L. Ultrafast and accurate 16S rRNA microbial community analysis using Kraken 2. Microbiome 8 , 124 (2020). https://doi.org/10.1186/s40168-020-00900-2 Dixon, P. VEGAN, a package of R functions for community ecology. Journal of vegetation science 14 , 927-930 (2003). https://doi.org/10.1111/j.1654-1103.2003.tb02228.x Seibenhener, M. L. & Wooten, M. C. Use of the Open Field Maze to measure locomotor and anxiety-like behavior in mice. J Vis Exp , e52434 (2015). https://doi.org/10.3791/52434 Chen, C. X. et al. A Multistep Workflow to Evaluate Newly Generated iPSCs and Their Ability to Generate Different Cell Types. Methods Protoc 4 (2021). https://doi.org/10.3390/mps4030050 Dorion, M.-F. et al. An adapted protocol to derive microglia from stem cells and its application in the study of CSF1R-related disorders. Molecular Neurodegeneration 19 , 31 (2024). https://doi.org/10.1186/s13024-024-00723-x Additional Declarations There is NO Competing Interest. Supplementary Files ArangurenExtendedDataTablev24.docx Extended Data Table 4 ArangurenExtendedDataTable3.docx Extended Data Table 3 ExtendedDataFigure2alphabetaFinal.pdf Extended Data Figure 2 ExtendedDataFigure4Final.pdf Extended Data Figure 4 ExtendedDataFigure5EVcharacteristicsFinal.pdf Extended Data Figure 5 ExtendedDataFigure8Final.pdf Extended Data Figure 8 ExtendedDataFigure1VisitComparisonFinal.pdf Extended Data Figure 1 ExtendedDataFigure6correlationFinal.pdf Extended Data Figure 6 ExtendedDataFigure9BiodistributionFinal.pdf Extended Data Figure 9 ArangurenExtendedDataTable2.docx Extended Data Table 2 ExtendedDataFigure7RNAseqFinal.pdf Extended Data Figure 7 ArangurenExtendedDataTable1.docx Extended Data Table 1 ExtendedDataFigure3FAMDFinal.pdf Extended Data Figure 3 ExtendedDatafigurelegends.docx Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8876425","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Biological Sciences - Article","associatedPublications":[],"authors":[{"id":595807335,"identity":"be099dde-9f93-42f9-bb93-fcf56bb2275b","order_by":0,"name":"Matheus Aranguren","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Matheus","middleName":"","lastName":"Aranguren","suffix":""},{"id":595807336,"identity":"27cb5360-41fb-4a91-9c0c-7396c82da632","order_by":1,"name":"Kim Doyon-Laliberté","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Kim","middleName":"","lastName":"Doyon-Laliberté","suffix":""},{"id":595807337,"identity":"32bc8444-8c23-4ea4-acfa-414a7bd8f90c","order_by":2,"name":"Idia Boncheva","email":"","orcid":"https://orcid.org/0009-0001-6485-8159","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Idia","middleName":"","lastName":"Boncheva","suffix":""},{"id":595807338,"identity":"59988d69-4b45-4322-8c84-673c69dd7744","order_by":3,"name":"Alexandre Villard","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Alexandre","middleName":"","lastName":"Villard","suffix":""},{"id":595807339,"identity":"2fb34232-1f95-47c1-81bb-63279057c3e7","order_by":4,"name":"Aléhandra Desjardins","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montré","correspondingAuthor":false,"prefix":"","firstName":"Aléhandra","middleName":"","lastName":"Desjardins","suffix":""},{"id":595807340,"identity":"1bbfbc72-fa26-48c4-9d7e-310a65b4f154","order_by":5,"name":"Emma Darbinian","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Emma","middleName":"","lastName":"Darbinian","suffix":""},{"id":595807341,"identity":"f12ba181-7988-45b8-a4b3-c3e3439fbfef","order_by":6,"name":"Suhani Patel","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Suhani","middleName":"","lastName":"Patel","suffix":""},{"id":595807342,"identity":"35769355-97f7-43e3-80aa-f193823d7eba","order_by":7,"name":"Charlotte DuSablon","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Charlotte","middleName":"","lastName":"DuSablon","suffix":""},{"id":595807343,"identity":"105cfd38-a0be-4a81-976d-1840f80b0dc5","order_by":8,"name":"Estefania Rivera Conde","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Estefania","middleName":"Rivera","lastName":"Conde","suffix":""},{"id":595807344,"identity":"b81ddfdb-6b74-4df2-be19-9a40d77d7bdd","order_by":9,"name":"Diana Cabrera Munoz","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Diana","middleName":"Cabrera","lastName":"Munoz","suffix":""},{"id":595807345,"identity":"1566fa6e-7ca0-4601-bf1c-fe8cea6f2f29","order_by":10,"name":"Ludhovik Purchase","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Ludhovik","middleName":"","lastName":"Purchase","suffix":""},{"id":595807346,"identity":"9b8f0828-5b9c-49a9-b1e9-731de3776ba2","order_by":11,"name":"Valerio Piscopo","email":"","orcid":"","institution":"McGill University","correspondingAuthor":false,"prefix":"","firstName":"Valerio","middleName":"","lastName":"Piscopo","suffix":""},{"id":595807347,"identity":"ca0c6f88-349e-41f5-811e-564d510b99a9","order_by":12,"name":"Aeshah Alluli","email":"","orcid":"","institution":"McGill University","correspondingAuthor":false,"prefix":"","firstName":"Aeshah","middleName":"","lastName":"Alluli","suffix":""},{"id":595807348,"identity":"33ded9ae-d113-4028-ab69-e14ad148c658","order_by":13,"name":"Faiza Benaliouad","email":"","orcid":"","institution":"McGill Univerity","correspondingAuthor":false,"prefix":"","firstName":"Faiza","middleName":"","lastName":"Benaliouad","suffix":""},{"id":595807349,"identity":"be4b2712-7d6c-4bab-9fe6-d2a1185def78","order_by":14,"name":"Julien Sirois","email":"","orcid":"","institution":"McGill University","correspondingAuthor":false,"prefix":"","firstName":"Julien","middleName":"","lastName":"Sirois","suffix":""},{"id":595807350,"identity":"48c3f2b2-42fe-4dd7-b363-7fa6673c7124","order_by":15,"name":"Thomas Durcan","email":"","orcid":"https://orcid.org/0000-0003-3942-1956","institution":"McGill University","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Durcan","suffix":""},{"id":595807351,"identity":"c61e77a3-0a46-4d6f-b300-603ed9002733","order_by":16,"name":"Chantal Massé","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Chantal","middleName":"","lastName":"Massé","suffix":""},{"id":595807352,"identity":"45506e54-78f4-41a9-b679-57d2868da971","order_by":17,"name":"Kodjovi Mlaga","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Kodjovi","middleName":"","lastName":"Mlaga","suffix":""},{"id":595807353,"identity":"91f8fb36-df5e-47ab-9955-7cb9b8bed520","order_by":18,"name":"Prabha Chandrasekaran","email":"","orcid":"","institution":"eHealth and Informatics; Caladrion","correspondingAuthor":false,"prefix":"","firstName":"Prabha","middleName":"","lastName":"Chandrasekaran","suffix":""},{"id":595807354,"identity":"976ce2a5-053c-43ab-aa6e-8cae85ccb129","order_by":19,"name":"Johanne Poudrier","email":"","orcid":"","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":false,"prefix":"","firstName":"Johanne","middleName":"","lastName":"Poudrier","suffix":""},{"id":595807334,"identity":"a75da003-d750-4a42-b779-566d095fc239","order_by":20,"name":"Emilia Falcone","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAklEQVRIie3SsUrDQBzH8V8JpMs/3Cb/0tK8QkIHl2BfJUHQqeDoEOSkUDfnQvsQ6RucBOJg1QfI0tC1szgVmwMJSo9mFLzv9j/y4X8HAWy2P1nnXsWAQFdPEbUhsiY96ejpqg0BFBqSn/5cnOVSbVKwmDrV9ub2fXD+8BJjl8IX8jjpLZLDxQow5+5oNF+XNFhPss6yQDhXx0lQ1sTFHXJy+96sJGYvczyJGAYy1mQP9jXZvzXEN23pH0gyAweaSNWQwEC43pI8Mof1W6i4JKZJ9rQsOFwZiFhcV9XnR8TD52m1pfRizN3X1WaXRv7QQL6X/RwV/T45Xbt/wGaz2f5JX9Y/VDMFQreTAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-5486-1549","institution":"Montreal Clinical Research Institute / Institut de recherches cliniques de Montréal (IRCM)","correspondingAuthor":true,"prefix":"","firstName":"Emilia","middleName":"","lastName":"Falcone","suffix":""}],"badges":[],"createdAt":"2026-02-14 02:55:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8876425/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8876425/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105538002,"identity":"60ca1ba6-ad9e-4888-8569-b65fcf937b90","added_by":"auto","created_at":"2026-03-27 07:36:39","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1174424,"visible":true,"origin":"","legend":"\u003cp\u003eSex-specific intestinal dysbiosis in Long COVID associates with neurological symptom burden and transfers barrier and neuroinflammatory phenotypes to gnotobiotic mice\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e IPCO cohort sub-study design and sample processing. Pandemic controls (PC; n=12) and participants with Long COVID (LC; n=91; not hospitalized during acute infection) provided stool samples for shotgun metagenomic sequencing at 3-6 months and/or 12 months after infection; a subset was used for gut microbiota-derived extracellular vesicle (GMEV) extraction. Created in BioRender. Falcone, E. (2026) https://BioRender.com/g6cp3ke. \u003cstrong\u003eb,\u003c/strong\u003e Forest plot showing odds ratios (ORs) for individual LC symptoms in females relative to males. Boxes indicate ORs and horizontal lines indicate 95% confidence intervals (log scale); symptom domains are descriptive groupings for visualization. \u003cstrong\u003ec\u003c/strong\u003e, Phylum-level community composition in females (top) and males (bottom) stratified by neurological symptom burden and sampling time point (3-6 months and 12 months after infection). \u003cstrong\u003ed\u003c/strong\u003e, Prevalence of selected dominant bacterial genera in females (left) and males (right) stratified by neurological symptom burden and sampling time point. \u003cstrong\u003ee\u003c/strong\u003e, Redundancy analysis (RDA) ordination of genus-level microbiome profiles in females (left) and males (right), with points representing samples coloured by neurological symptom group and sampling time point; arrows (biplot vectors) indicate taxa and symptom loadings and their direction of association with the ordination axes. \u003cstrong\u003ef\u003c/strong\u003e, Experimental design for faecal microbiota transfer into germ-free (GF) mice using stool from female LC donors with neurological symptoms (LC Neuro; n=4 donors) or without neurological symptoms (LC No-Neuro; n=4 donors). Created in BioRender. Aranguren, M. (2026) https://BioRender.com/8tx74nl. \u003cstrong\u003eg\u003c/strong\u003e, Open-field test in GF mice colonised with LC No-Neuro or LC Neuro donor microbiota; mean velocity (left) and total distance travelled (right) are shown (each dot represents one mouse; n=5 mice per group). \u003cstrong\u003eh\u003c/strong\u003e, Representative immunofluorescence images of glial fibrillary acidic protein (GFAP; astrocytes; top) and ionized calcium-binding adapter molecule 1 (IBA1; microglia; bottom) in brain sections from GF mice colonised with LC No-Neuro or LC Neuro donor microbiota (20×). Scale bar, 100 µm. \u003cstrong\u003ei\u003c/strong\u003e, Quantification of GFAP and IBA1 fluorescence intensity density. Each dot represents one field (5 fields per mouse; n=3 mice per donor group).\u003c/p\u003e","description":"","filename":"fig1.png","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/1e90afba787645f923e59e62.png"},{"id":105538004,"identity":"e59b153d-5e71-4058-acb3-9f280482ac1b","added_by":"auto","created_at":"2026-03-27 07:36:39","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":611625,"visible":true,"origin":"","legend":"\u003cp\u003eGut microbiota-derived extracellular vesicles from individuals with long COVID induce epithelial inflammasome signalling, impair barrier function and activate macrophages in vitro.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ea,\u003c/strong\u003e Human induced pluripotent stem cell-derived intestinal organoid (HIO) monolayers were treated with vehicle (PBS; n=3) or gut microbiota-derived extracellular vesicles (GMEVs; 1μg/mL) isolated from pandemic controls (PC; n=6 donors), long COVID (LC) donors without neurological symptoms (LC No-Neuro; n=10 donors) or LC donors with neurological symptoms (LC Neuro; n=15 donors). HIO transcript levels (including \u003cem\u003eNLRP3, IL1B, TNFSF13B\u003c/em\u003e) were quantified by qRT-PCR at 5h, and secreted mediators (including IL-1b, IL-6, IL-8, TNF, CCL2 and CXCL10) were quantified at 18h using Meso Scale Discovery (MSD) assays. \u003cstrong\u003eb, \u003c/strong\u003eTransepithelial electrical resistance (TEER) across Caco-2 monolayers measured 24h after treatment with GMEVs (3μg/mL); values are shown as post/pre ratio. \u003cstrong\u003ec,\u003c/strong\u003e THP-1-derived macrophages treated with vehicle (PBS; n=3) or donor-derived GMEVs (PC, n=6; LC No-Neuro, n=10; LC Neuro, n=15). Intracellular IL-1b and TNF frequencies and BAFF-associated responses were quantified by flow cytometry after 18h stimulation with brefeldin A added for the final 5h; macrophage inflammatory transcripts (including \u003cem\u003eNLRP3\u003c/em\u003e, \u003cem\u003eTNF\u003c/em\u003eand \u003cem\u003eTNFSF13B\u003c/em\u003e) were measured by qRT–PCR at 5h. \u003cstrong\u003ed\u003c/strong\u003e, Pairwise correlation matrices of macrophage cytokine and chemokine responses across conditions (PC, LC No-Neuro and LC Neuro). Asterisks denote significant correlations (*\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05, **\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.01, ***\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.001, ****\u003cem\u003eP\u003c/em\u003e \u0026lt; 0.0001). In \u003cstrong\u003ea–c\u003c/strong\u003e, each dot represents an independent donor-derived GMEV preparation (or an independent experiment for vehicle controls); centre lines indicate medians and boxes indicate interquartile ranges.\u003c/p\u003e","description":"","filename":"fig2.png","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/b160d81776e445b0c54388bd.png"},{"id":105567293,"identity":"9c191f8e-da8d-4716-968c-a4b05c53642f","added_by":"auto","created_at":"2026-03-27 12:58:47","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":589294,"visible":true,"origin":"","legend":"\u003cp\u003eGut microbiota-derived extracellular vesicles from individuals with long COVID induce a neuroinflammatory transcriptional programme in induced pluripotent stem cell-derived microglia\u003c/p\u003e\n\u003cp\u003eInducted pluripotent stem cell-derived microglia (iMGLs) were exposed for 16h to gut microbiota-derived extracellular vesicles (GMEVs; 1µg/mL) from long COVID donors (LC; n=10), pandemic controls (PC; n=4), or vehicle (PBS; n=3), followed by bulk RNA sequencing. \u003cstrong\u003ea\u003c/strong\u003e, Principal component analysis (PCA) of transcriptomes from GMEV-treated iMGLs, showing separation of LC and PC conditions along the principal axes. \u003cstrong\u003eb\u003c/strong\u003e, Volcano plot of differentially expressed genes in LC versus PC conditions (adjusted P \u0026lt; 0.05). Selected significantly regulated genes are annotated. \u003cstrong\u003ec\u003c/strong\u003e, Gene Ontology (GO) circle plot summarizing enriched biological process terms among differentially expressed genes. Bar height indicates significance (−log10 adjusted \u003cem\u003eP\u003c/em\u003e value), colour indicates Z-score (directionality), and scatter points represent individual genes plotted by log2 fold change. \u003cstrong\u003ed\u003c/strong\u003e, Top 20 enriched KEGG pathways ranked by −log10 adjusted \u003cem\u003eP\u003c/em\u003e value. \u003cstrong\u003ee\u003c/strong\u003e, Reactome pathway enrichment analysis highlighting immune and cytokine signalling pathways among LC-induced genes. \u003cstrong\u003ef\u003c/strong\u003e, Heatmap of genes within the KEGG axon guidance pathway (hsa04360). Rows represent genes and columns represent samples; red indicates upregulation and blue indicates downregulation (Z-score scaled). \u003cstrong\u003eg\u003c/strong\u003e, Quantitative RT–PCR validation of selected inflammatory and neuroimmune-associated transcripts in iMGLs treated with vehicle, PC-derived GMEVs or LC-derived GMEVs. \u003cstrong\u003eh\u003c/strong\u003e, Cytokine and chemokine secretion measured by Meso Scale Discovery (MSD) assays following GMEV stimulation. In \u003cstrong\u003eg,h\u003c/strong\u003e, each dot represents one donor-derived GMEV preparation (LC or PC) or one independent vehicle experiment; centre lines indicate medians and boxes indicate interquartile ranges.\u003c/p\u003e","description":"","filename":"fig3.png","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/3488caed564e5b6a84eae67f.png"},{"id":105538016,"identity":"c6d5372d-825b-4b4e-99d4-fa8eda16d397","added_by":"auto","created_at":"2026-03-27 07:36:40","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":216323,"visible":true,"origin":"","legend":"\u003cp\u003eOral administration of gut microbiota-derived extracellular vesicles from individuals with long COVID remodels the faecal microbiota of wild-type mice\u003c/p\u003e\n\u003cp\u003eFemale wild-type mice (6-8 weeks old) received oral gavage of vehicle (PBS; n=9) or gut microbiota-derived extracellular vesicles (GMEVs; 5µg per dose) isolated from pandemic controls (PC; 3 GMEV donors; n=4 mice per GMEV donor), LC donors without neurological symptoms (LC No-Neuro; GMEV donors; n=4 mice per GMEV donor) or LC donors with neurological symptoms (LC Neuro; GMEV donors; n=4 mice per GMEV donor). GMEVs were administered 3 times per week for 6 weeks, after which faecal microbiota composition was profiled by 16S rRNA gene sequencing. The experiment was performed in 3 independent runs, each using an independent donor-derived GMEV preparation per group. \u003cstrong\u003ea\u003c/strong\u003e, Alpha diversity (Shannon diversity and Pielou’s evenness). \u003cem\u003eP\u003c/em\u003e values were calculated by Kruskal-Wallis test. \u003cstrong\u003eb\u003c/strong\u003e, Relative abundance of bacterial phyla across groups. \u003cstrong\u003ec\u003c/strong\u003e, Principal coordinates analysis (PCoA) of Bray-Curtis dissimilarity showing separation of faecal communities by treatment group. \u003cstrong\u003ed\u003c/strong\u003e, Cladogram from linear discriminant analysis effect size (LEfSe) showing taxa that differ in relative abundance across groups (linear discriminant analysis (LDA) score ≥ 2; \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05). \u003cstrong\u003ee\u003c/strong\u003e, LEfSe LDA scores for differentially abundant genera; bars indicate effect size and points indicate relative abundance.\u003c/p\u003e","description":"","filename":"fig4.png","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/c571b0534b7f9d34eaf0c0a5.png"},{"id":105566903,"identity":"4b48d676-1f2e-4410-a4f0-850c7edd7725","added_by":"auto","created_at":"2026-03-27 12:57:40","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1919458,"visible":true,"origin":"","legend":"\u003cp\u003eOral administration of gut microbiota-derived extracellular vesicles from individuals with long COVID induces intestinal inflammatory signatures, behavioural alterations and glial activation in wild-type mice\u003c/p\u003e\n\u003cp\u003eFemale C57BL/6 mice (6-8 weeks old) received oral gavage of gut microbiota–derived extracellular vesicles (GMEVs; 5 µg per dose) isolated from LC donors without neurological symptoms (No-Neuro; 3 GMEV donors; n=4 mice per GMEV donor), LC donors with neurological symptoms (Neuro; 3 GMEV donors; n=4 mice per GMEV donor) or pandemic controls (PC; 3 GMEV donors; n=4 mice per GMEV donor), or vehicle (PBS; n=9 mice), 3 times per week for 6 weeks. The experiment was performed in 3 independent runs using one donor-derived GMEV preparation per group per run. \u003cstrong\u003ea\u003c/strong\u003e, Experimental design. Created in BioRender. Falcone, E. (2026) https://BioRender.com/nk3kim1. \u003cstrong\u003eb\u003c/strong\u003e, Colon length and serum levels of lipopolysaccharide-binding protein (LBP) and B-cell activating factor (BAFF) at endpoint.\u0026nbsp;\u003cstrong\u003ec\u003c/strong\u003e, Colonic epithelial expression of \u003cem\u003eNlrp3\u003c/em\u003e, \u003cem\u003eIl1b\u003c/em\u003e, \u003cem\u003eTnfsf13b\u003c/em\u003e, \u003cem\u003eTnf\u003c/em\u003e and \u003cem\u003eIl6\u003c/em\u003e measured by qRT–PCR.\u0026nbsp;\u003cstrong\u003ed\u003c/strong\u003e, Representative immunofluorescence images of zonula occludens 1 (ZO-1) in colonic sections (green) with DAPI nuclear counterstain (blue); dashed boxes indicate regions shown at higher magnification (40´).\u0026nbsp;Scale bar, 100 µm. Representative images are shown from n=2 mice per group. \u003cstrong\u003ee\u003c/strong\u003e, Open-field test, including distance travelled, velocity, time spent moving versus not moving and centre-border exploration metrics. \u003cstrong\u003ef\u003c/strong\u003e, Y-maze spontaneous alternation behaviour, including distance travelled, velocity, total alternations, proportion of successful alternations and indirect revisits. Each dot represents one mouse; violin plots show the distribution, centre lines denote medians and boxes indicate interquartile ranges. \u003cstrong\u003eg,h\u003c/strong\u003e, Representative immunofluorescence images of glial fibrillary acidic protein (GFAP; astrocytes; \u003cstrong\u003eg\u003c/strong\u003e) and ionized calcium-binding adapter molecule 1 (IBA1; microglia; \u003cstrong\u003eh\u003c/strong\u003e) in brain sections, with DAPI counterstain. \u003cstrong\u003ei\u003c/strong\u003e, Quantification of GFAP and IBA1 fluorescence intensity density. Each dot represents one field (5 fields per mouse; n=3 mice per group).\u003c/p\u003e","description":"","filename":"fig5.png","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/69e5b9dc6790cc87128f840c.png"},{"id":105728390,"identity":"7f20fd29-3039-4fcb-9c13-d17b60569cd5","added_by":"auto","created_at":"2026-03-30 11:11:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5324592,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/6a0664c9-322d-431b-979b-2fea09cdbb04.pdf"},{"id":105566775,"identity":"5b9c43ed-8890-4132-8fc9-ddb482a20d0b","added_by":"auto","created_at":"2026-03-27 12:57:17","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":17237,"visible":true,"origin":"","legend":"Extended Data Table 4","description":"","filename":"ArangurenExtendedDataTablev24.docx","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/15a61ebb235d456659694db2.docx"},{"id":105567915,"identity":"fdd08483-d1db-48cb-96af-b4ced841cdc8","added_by":"auto","created_at":"2026-03-27 13:05:58","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":18594,"visible":true,"origin":"","legend":"Extended Data Table 3","description":"","filename":"ArangurenExtendedDataTable3.docx","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/3a80b026a03a039fb9ad44d0.docx"},{"id":105538009,"identity":"327dbc70-8409-4cca-9309-9456f01ce459","added_by":"auto","created_at":"2026-03-27 07:36:39","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":259706,"visible":true,"origin":"","legend":"Extended Data Figure 2","description":"","filename":"ExtendedDataFigure2alphabetaFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/a399f835b233c628799d2b27.pdf"},{"id":105538019,"identity":"d291f5e1-3224-4b22-8faf-08353885cf69","added_by":"auto","created_at":"2026-03-27 07:36:40","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":1084743,"visible":true,"origin":"","legend":"Extended Data Figure 4","description":"","filename":"ExtendedDataFigure4Final.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/e795d4a79116306e9a1c7d22.pdf"},{"id":105538014,"identity":"accd3253-6ad3-44a9-9757-54f3db3fd164","added_by":"auto","created_at":"2026-03-27 07:36:40","extension":"pdf","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":1665940,"visible":true,"origin":"","legend":"Extended Data Figure 5","description":"","filename":"ExtendedDataFigure5EVcharacteristicsFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/bb1c7c7a5d4891f72ad2f0b6.pdf"},{"id":105567347,"identity":"0f11f55a-978b-4ba8-990b-6cee45ce2d16","added_by":"auto","created_at":"2026-03-27 12:59:02","extension":"pdf","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":3748027,"visible":true,"origin":"","legend":"Extended Data Figure 8","description":"","filename":"ExtendedDataFigure8Final.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/96f4f517fe56c3057beaf921.pdf"},{"id":105567304,"identity":"8abf0886-1c24-4504-a19d-6650ff1969f9","added_by":"auto","created_at":"2026-03-27 12:58:49","extension":"pdf","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":2258364,"visible":true,"origin":"","legend":"Extended Data Figure 1","description":"","filename":"ExtendedDataFigure1VisitComparisonFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/8806a34bcaaabaef9d9cfdad.pdf"},{"id":105567396,"identity":"83192246-0b99-46dd-bb25-1e9273a13c36","added_by":"auto","created_at":"2026-03-27 12:59:18","extension":"pdf","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":184841,"visible":true,"origin":"","legend":"Extended Data Figure 6","description":"","filename":"ExtendedDataFigure6correlationFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/5d8dc2f59029a6a4a6d622ee.pdf"},{"id":105538013,"identity":"8dbc323b-c9a9-485c-8282-59a78a34c9d6","added_by":"auto","created_at":"2026-03-27 07:36:39","extension":"pdf","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":2854384,"visible":true,"origin":"","legend":"Extended Data Figure 9","description":"","filename":"ExtendedDataFigure9BiodistributionFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/489b0f157a8307a73c0a912b.pdf"},{"id":105538018,"identity":"e0104c67-7ba5-4293-ba84-e4f02eda5f07","added_by":"auto","created_at":"2026-03-27 07:36:40","extension":"docx","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":23934,"visible":true,"origin":"","legend":"Extended Data Table 2","description":"","filename":"ArangurenExtendedDataTable2.docx","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/776ea29a319fa7bb8678a3d7.docx"},{"id":105566996,"identity":"351f48a8-b4b3-43c7-a01a-5120b91b342a","added_by":"auto","created_at":"2026-03-27 12:57:58","extension":"pdf","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":6339416,"visible":true,"origin":"","legend":"Extended Data Figure 7","description":"","filename":"ExtendedDataFigure7RNAseqFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/3d8d706777fb53fea4a99eed.pdf"},{"id":105567430,"identity":"fa611a0b-dba8-43ab-97af-9976dd1ee108","added_by":"auto","created_at":"2026-03-27 12:59:23","extension":"docx","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":19056,"visible":true,"origin":"","legend":"Extended Data Table 1","description":"","filename":"ArangurenExtendedDataTable1.docx","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/649048ca295bf3d997094533.docx"},{"id":105538020,"identity":"9b3963c3-9a36-4d53-bb72-9f967a9de50c","added_by":"auto","created_at":"2026-03-27 07:36:40","extension":"pdf","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":25420239,"visible":true,"origin":"","legend":"Extended Data Figure 3","description":"","filename":"ExtendedDataFigure3FAMDFinal.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/25b825bbc634d52c59ade503.pdf"},{"id":105566864,"identity":"5d1c9ca7-1768-4e86-b246-943e42183f40","added_by":"auto","created_at":"2026-03-27 12:57:35","extension":"docx","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":17889,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDatafigurelegends.docx","url":"https://assets-eu.researchsquare.com/files/rs-8876425/v1/9bd40a64ba19bd3c6cc61c97.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Microbiota-derived extracellular vesicles mediate gut-brain axis dysfunction in long COVID","fulltext":[{"header":"Introduction","content":"\u003cp\u003eApproximately 5-19% of individuals infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) develop post COVID-19 condition (PCC), also referred to as post-acute sequelae of COVID-19 (PASC) or Long COVID (LC)\u003csup\u003e2,5\u003c/sup\u003e. LC is a post-infectious chronic condition present for at least 3 months after acute SARS-CoV-2 infection, and it may follow a continuous, relapsing-remitting or progressive disease course affecting one or multiple organ systems\u003csup\u003e2,6\u003c/sup\u003e. Although more than 200 symptoms have been attributed to LC\u003csup\u003e7,8\u003c/sup\u003e, some of the most prevalent and debilitating manifestations include fatigue, post-exertional malaise, cognitive impairment (\u0026ldquo;brain fog\u0026rdquo;), memory and concentration difficulties, dyspnoea and cardiac \u0026nbsp;dysautonomia, including postural orthostatic tachycardia syndrome\u003csup\u003e4,9,10\u003c/sup\u003e. These sequelae can substantially impair functional capacity and quality of life for years after infection\u003csup\u003e2,11\u003c/sup\u003e. Despite its growing public health burden, LC currently lacks validated diagnostic biomarkers and targeted pharmacological therapies. Accumulating evidence implicates immune \u0026nbsp;dysregulation\u003csup\u003e12-14\u003c/sup\u003e, chronic systemic inflammation\u003csup\u003e15,16\u003c/sup\u003e,\u0026nbsp;microclotting\u003csup\u003e17\u003c/sup\u003e,\u0026nbsp;neuroinflammation\u003csup\u003e18,19\u003c/sup\u003e, autoimmunity\u003csup\u003e20,21\u003c/sup\u003e and viral persistence\u003csup\u003e22,23\u003c/sup\u003e, including prolonged viral shedding in the gastrointestinal (GI) tract\u003csup\u003e24-26\u003c/sup\u003e, as features of LC pathophysiology. \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAcute SARS-CoV-2 infection is associated with intestinal dysbiosis, particularly in individuals with severe disease\u003csup\u003e27\u003c/sup\u003e, and alterations in the gut microbiota can persist for at least 6 months in individuals with LC\u003csup\u003e28\u003c/sup\u003e. However, the mechanisms by which SARS-CoV-2-associated dysbiosis contributes to immune dysregulation and end-organ injury, including neuroinflammation, remain poorly defined. The intestinal microbiota plays a central role in immune homeostasis\u003csup\u003e29\u003c/sup\u003e, particularly at the mucosal barrier\u003csup\u003e30\u003c/sup\u003e,\u0026nbsp;and dysbiosis has been linked to increased intestinal inflammation and epithelial permeability (\u0026ldquo;leaky gut\u0026rdquo;), permitting the translocation of microbial products into systemic circulation\u003csup\u003e31\u003c/sup\u003e. Increased intestinal permeability has been documented during acute COVID-19 infection\u003csup\u003e32,33\u003c/sup\u003e, and circulating markers of microbial translocation are elevated in individuals with LC\u003csup\u003e34\u003c/sup\u003e. Although the gut-brain axis has been implicated in neuropsychological and neurodegenerative disorders\u003csup\u003e35\u003c/sup\u003e, including in chronic viral infections such as human immunodeficiency virus (HIV)\u003csup\u003e36-38\u003c/sup\u003e, a causal relationship between intestinal dysbiosis, barrier dysfunction, immune activation, and neuroinflammation has not been demonstrated in LC. \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp; \u0026nbsp;\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBacteria communicate with each other and their host through the release of extracellular vesicles (EVs), nanoscale (20-400 nm) lipid bilayer particles that transport proteins, lipids, metabolites and nucleic acids\u003csup\u003e39\u003c/sup\u003e. EVs produced by the intestinal microbiota, here termed gut microbiota-derived extracellular vesicles (GMEVs), can cross the intestinal barrier, enter the bloodstream and act on distant tissues, thereby contributing to diverse inflammatory and metabolic disorders, including metabolic syndrome and liver disease\u003csup\u003e40-42\u003c/sup\u003e. We hypothesized that GMEVs produced by the perturbed microbiota of individuals with LC promote intestinal barrier dysfunction, systemic inflammation and downstream neuroinflammation. Here, we functionally characterized the faecal microbiome of individuals with LC at 3-6 and 12 months post-infection and examined the biological effects of LC-derived GMEVs using complementary in vitro and in vivo\u003cem\u003e\u0026nbsp;\u003c/em\u003emodels. Our findings identify GMEVs as mechanistic mediators linking intestinal dysbiosis to barrier dysfunction, immune activation, and neuropsychological manifestations in LC.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eWomen with LC and neurological symptoms have intestinal dysbiosis that transfers neurobehavioral alteration and neuroinflammation to gnotobiotic mice\u003c/p\u003e\n\u003cp\u003eWe analysed a sub-cohort of 103 participants from the Institut de recherches cliniques de Montr\u0026eacute;al (IRCM) Post-COVID-19 (IPCO) cohort (NCT04736732), comprising 12 pandemic controls (PC; individuals who never reported COVID-19 symptoms or tested positive for SARS-CoV-2) and 91 individuals with LC who were not hospitalized during acute infection, all of whom provided stool samples for shotgun metagenomic sequencing (Fig.1a). The median age (IQR) of PC and LC participants was 50 (46.5-63) and 47 (39-58) years, respectively, and 41.7% of PC and 64.8% of LC participants self-identified as female. Female sex was associated with higher odds of LC symptom reporting overall (odds ratio (OR)=1.26, p=6.3 x 10\u003csup\u003e-9\u003c/sup\u003e; Fig.1b) including several neurological symptoms, such as trouble with memory (OR=12.2, \u003cem\u003eP\u003c/em\u003e=0.018), confusion (OR=91.95, \u003cem\u003eP\u003c/em\u003e=0.043), vertigo or dizziness (OR=12.44, \u003cem\u003eP\u003c/em\u003e=0.042) and nausea (OR=2183, \u003cem\u003eP\u003c/em\u003e=0.022).\u003c/p\u003e\n\u003cp\u003eAcross sexes, LC participants exhibited intestinal microbiome profiles that were distinct from PC participants at 3-6 months and 12 months following acute infection, with differences in alpha diversity, beta diversity and taxonomic features that persisted over time (Extended Data Fig.1). To assess whether neurological symptom burden tracked with specific microbiome features, LC participants were stratified by neurological symptom burden. Female LC participants were categorized as having severe neurological symptoms (Severe Neuro; \u0026sup3;3 of 8 assessed neurocognitive /neuropsychological symptoms), mild neurological symptoms (Mild Neuro; \u0026lt;3 symptoms), or no neurological symptoms (No-Neuro) (Fig.1b). In females, Severe Neuro LC participants showed higher microbial diversity and evenness than No Neuro LC participants (Extended Data Fig.2a), together with shifts in community composition (Extended Data Fig.2b). A similar pattern was observed in males, with neurological symptom burden associated with differences in diversity and beta diversity at 6 and/or 12 months (Extended Data Fig.2c,d). Factor analysis of mixed data (FAMD) further supported an association of neurological and respiratory symptom dimensions with microbiome composition in both sexes (Extended Data Fig.3). At the phylum level, community composition differed across neurological symptom strata and timepoints in both females and males (Fig.1c).\u003c/p\u003e\n\u003cp\u003eTaxonomic analyses identified sex-specific patterns of dysbiosis associated with neurological symptom burden. In females, Severe Neuro LC participants showed depletion in multiple genera within the class \u003cem\u003eClostridia\u003c/em\u003e, including \u003cem\u003eRuminococcus\u003c/em\u003e, \u003cem\u003eLachnospiraceae,\u0026nbsp;\u003c/em\u003eand \u003cem\u003eFusicatenibacter,\u003c/em\u003e alongside enrichment of general within order \u003cem\u003eBacteroidales\u003c/em\u003e, including \u003cem\u003eBacteroides\u003c/em\u003e, \u003cem\u003ePhocaeicola\u003c/em\u003e, and \u003cem\u003eAlistipes\u003c/em\u003e, relative to Mild Neuro and No-Neuro groups; these differences were maintained through 12 months (Fig.1d, left). In males, Severe and Mild Neuro LC participants showed depletion of \u003cem\u003eAlistipes\u003c/em\u003e, \u003cem\u003eRoseburia\u003c/em\u003e and \u003cem\u003eBifidobacterium\u003c/em\u003e relative to No-Neuro LC participants at both 6 and 12 months, with additional genus-level differences emerging at 12 months (Fig.1d, right). Ordination analyses linked these taxa shifts to specific neurological symptoms, including trouble with memory and concentration, confusion and brain fog, with stronger symptom-taxa associations in females (Fig.1e).\u003c/p\u003e\n\u003cp\u003eGiven the higher burden of neurological symptoms among female participants, we tested whether the associated microbiota features were functionally transferable. Germ-free female mice were colonised with stool from female LC donors with neurological symptoms (LC Neuro; n=3 donors) or from female LC donors without neurological symptoms (LC No Neuro; n=3 donors) (Fig.1f). Mice colonized with LC Neuro microbiota showed evidence of impaired intestinal barrier integrity, including reduced/disrupted zonula occludens 1 (ZO-1) staining in colonic tissue (Extended Data Fig.4b), and displayed neurobehavioral differences, including increased locomotion activity in the open-field test (Fig.1g, Extended Data Fig.4a). These mice also developed neuroinflammatory changes, including accumulation of activated astrocytes in the hippocampal region and amoeboid-shaped activated microglia in the hindbrain (Fig.1h,i). In contrast, mice colonized with LC No-Neuro donor microbiota did not show these pathological features. Together, these data indicate that intestinal microbiota associated with neurological symptoms in female LC participants is sufficient to induce intestinal barrier dysfunction, neurobehavioral alteration and neuroinflammatory phenotypes in gnotobiotic mice.\u003c/p\u003e\n\u003cp\u003eGMEVs from individuals with LC induce intestinal epithelial inflammation, impair barrier function and activate macrophages in vitro\u003c/p\u003e\n\u003cp\u003eTo test whether GMEVs contribute to LC-associated mucosal and systemic inflammation, we isolated GMEVs from stool pf LC participants with neurological symptoms (LC Neuro), LC participants without neurological symptoms (LC No-Neuro), and pandemic controls (PC). Transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA) confirmed vesicular morphology and broadly similar size distribution across groups, with comparable EV yield metrics as summarized in Extended Data Fig.5.\u003c/p\u003e\n\u003cp\u003eWe first assessed epithelial responses to GMEVs using induced pluripotent stem cell\u0026ndash;derived human intestinal organoid (HIO) monolayers from healthy donors. Exposure to GMEVs (1 \u0026mu;g/mL) increased inflammasome-associated transcripts in HIOs, including \u003cem\u003eNLRP3\u003c/em\u003e and \u003cem\u003eIL1B\u003c/em\u003e, relative to vehicle and PC-derived GMEVs, and also increased \u003cem\u003eTNFSF13B\u0026nbsp;\u003c/em\u003e(encoding B-cell activating factor (BAFF)) (Fig.2a). At the protein level, LC-derived GMEVs promoted secretion of inflammatory mediators, with the most pronounced increases generally observed following exposure to LC Neuro GMEVs across measured cytokines and chemokines (including IL-1\u0026beta;, IL-8, TNF, CCL2 and CXCL10) (Fig.2a).\u003c/p\u003e\n\u003cp\u003eIncreased intestinal permeability and microbial translocation are hallmarks of chronic intestinal inflammation\u003csup\u003e22,24\u003c/sup\u003e, and individuals with LC exhibit elevated circulating markers of barrier dysfunction and microbial translocation, including soluble zonulin, lipopolysaccharide-binding protein (LBP) and \u0026beta;-D-glucan\u003csup\u003e34,43\u003c/sup\u003e. We therefore evaluated whether GMEVs directly alter epithelial barrier integrity by measuring trans-epithelial electrical resistance (TEER) across Caco-2 cell monolayers. LC Neuro GMEVs produced the strongest reduction in TEER (post-pre), consistent with impaired barrier function, whereas LC No-Neuro GMEVs exerted more modest effects and PC-derived GMEVs had minimal impact (Fig.2b).\u003c/p\u003e\n\u003cp\u003eWe next asked whether GMEVs promote innate immune activation that could plausibly contribute to systemic inflammation. THP-1-derived macrophages responded to GMEVs from all donor groups compared to vehicle, consistent with a conserved immunostimulatory capacity of microbiota-derived vesicles (Fig.2c). However, LC Neuro GMEVs elicited the most robust macrophage activation, with higher frequencies and/or intensities of IL-1\u0026beta;-, TNF- and BAFF-associated responses compared with LC No-Neuro and PC conditions (Fig.2c; Extended Data Fig.6d). In parallel, LC-derived GMEVs increased expression of inflammatory transcripts in macrophages, including \u003cem\u003eNLRP3\u003c/em\u003e, \u003cem\u003eTNF\u003c/em\u003e and \u003cem\u003eTNFSF13B\u003c/em\u003e, with the strongest responses again observed in the LC Neuro condition (Fig.2c).\u003c/p\u003e\n\u003cp\u003eFinally, to compare inflammatory network structure across donor groups, we examined cytokine/chemokine relationships induced by GMEVs. Correlation analyses revealed a denser pattern of coordinated mediator responses in macrophages stimulated with LC Neuro GMEVs than with LC No-Neuro or PC GMEVs, including stronger coupling between \u003cem\u003eTNFSF13B\u003c/em\u003e and multiple inflammatory readouts (Extended Data Fig.6a\u0026ndash;c).\u003c/p\u003e\n\u003cp\u003eTogether, these data indicate that LC-derived GMEVs are sufficient to trigger epithelial inflammatory programmes and macrophage activation in vitro, and that GMEVs from LC Neuro donors most consistently associate with barrier-disruptive activity and coordinated multi-cytokine inflammatory responses.\u003c/p\u003e\n\u003cp\u003eExtracellular vesicles derived from the gut microbiota of individuals with LC induce a pro-inflammatory response in iPSC-derived microglia\u003c/p\u003e\n\u003cp\u003eBecause transplantation of LC-associated microbiota induced neuroinflammatory phenotypes in germ-free mice and LC-derived GMEVs activated macrophages in vitro, we asked whether GMEVs can also directly modulate central nervous system innate immune cells. We therefore exposed induced pluripotent stem cell\u0026ndash;derived microglia (iMGLs) to donor-derived GMEVs from LC participants (LC Neuro and LC No-Neuro) or pandemic controls (PC) and profiled transcriptional responses by RNA sequencing. Global ordination separated iMGLs stimulated with LC-derived GMEVs from those treated with PC-derived GMEVs, with LC Neuro and LC No-Neuro conditions clustering together, indicating broadly similar microglial responses across LC donor subgroups (Fig.3a). Differential expression analysis identified 2,631 genes altered in iMGLs exposed to LC-derived versus PC-derived GMEVs (adjusted \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05) (Fig.3b). Gene ontology enrichment highlighted a coordinated activation of immune and inflammatory processes, including innate immune response programmes, chemotaxis-related terms and cytokine-linked pathways (Fig.3c; Extended Data Fig.7a). Consistent with this, pathway-level analyses identified significant enrichment of immune and infection -associated signatures, ranked among the top differentially regulated pathways, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) \u0026ldquo;COVID-19\u0026rdquo; pathway (Fig.3d; Extended Data Fig.7b,c). Gene set enrichment analysis further showed preferential induction of inflammatory response modules in LC-GMEV-treated iMGLs, including interferon- and cytokine-associated signalling and neutrophil-linked programmes (Fig.3e). In addition, enrichment analyses identified pathways with established roles in neuroimmune communication, including axon guidance, which showed broad upregulation of constituent genes in iMGLs treated with LC-derived GMEVs (Fig.3f). Protein\u0026ndash;protein interaction network analysis of highly induced genes revealed interconnected modules centered on inflammatory and myeloid activation nodes, including networks annotated for regulation of neuroinflammatory responses (Extended Data Fig.8a).\u003c/p\u003e\n\u003cp\u003eWe validated key transcriptomic changes by targeted qPCR, confirming induction of inflammatory and myeloid-associated genes (for example, \u003cem\u003eIL1B\u003c/em\u003e, \u003cem\u003eTNF\u003c/em\u003e, \u003cem\u003eIL6\u003c/em\u003e, \u003cem\u003eAIF1\u003c/em\u003e and neutrophil-linked S100 family transcripts) in iMGLs exposed to LC-derived GMEVs compared with PC-derived GMEVs and vehicle (Fig.3g; Extended Data Fig.8b). At the protein level, LC-derived GMEVs increased secretion of inflammatory mediators, including IL-6, TNF and CXCL1 (Fig.3h). Notably, \u003cem\u003eTNFSF13B\u003c/em\u003e induction and increased BAFF protein were also observed, paralleling epithelial and macrophage responses and identifying BAFF as a convergent inflammatory mediator in response to GMEVs across multiple cellular compartments (Fig.3g,h).\u003c/p\u003e\n\u003cp\u003eTogether, these data show that GMEVs from individuals with LC, irrespective of neurological symptom subgroup, are sufficient to elicit a robust human microglial activation state marked by coordinated induction of inflammatory and neuroimmune-associated gene programmes, thereby supporting a role for LC-associated GMEVs in shaping microglial responses that may contribute to neuroinflammation in LC.\u003c/p\u003e\n\u003cp\u003eOral administration of GMEVs from individuals with LC remodels the intestinal microbiota in wild-type mice\u003c/p\u003e\n\u003cp\u003eWe next asked whether GMEVs are sufficient to reshape intestinal microbial communities in vivo. Wild-type (WT) mice received oral gavage of donor-derived GMEVs (5\u0026mu;g/mL) from LC Neuro participants, LC No-Neuro participants or PCs, or vehicle alone, 3 times per week for 6 weeks. To assess whether orally administered vesicles disseminate beyond the gut, we tracked fluorescently labelled bacterial EVs (derived from \u003cem\u003eEscherichia coli\u003c/em\u003e as a surrogate for GMEVs). Eight hours after gavage, fluorescent signal was detected in the gastrointestinal tract and in peripheral organs, including liver and kidney, and was also measurable in brain tissue, whereas signal was not detected in mice receiving unlabelled vesicles (Extended Data Fig.9).\u003c/p\u003e\n\u003cp\u003eTo determine whether GMEVs alter intestinal community structure, we performed 16S rRNA gene sequencing on faecal samples collected after 6 weeks of treatment. Relative to vehicle and PC groups, administration of GMEVs from LC donors, irrespective of neurological symptom status, was associated with reduced alpha diversity, with decreases in Shannon diversity and Pielou\u0026rsquo;s evenness indices (Fig.4a). At the phylum level, mice receiving GMEVs from LC Neuro donors showed the most pronounced compositional shifts, including altered relative abundance of Bacillota to Bacteroidota and reductions in Actinomycetota and Verrucomicrobiota compared to controls groups (Fig.4b). Principal coordinate analysis based on Bray-Curtis dissimilarity showed clear separation of microbiota profiles in mice treated with LC-derived GMEVs from those receiving vehicle or PC-derived GMEVs (Fig.4c). Notably, mice administered GMEVs from LC No-Neuro donors displayed a bifurcated clustering pattern, with partial overlap with both control and LC Neuro clusters, consistent with heterogenous responses to LC No-Neuro GMEVs.\u003c/p\u003e\n\u003cp\u003eWe next identified taxa contributing to these treatment-associated differences. Linear discriminant analysis effect size (LEfSe) revealed distinct microbial features associated with each treatment group (Fig.4d). Mice receiving LC Neuro-derived GMEVs were enriched for taxa within the family Lachinospiraceae and the class \u003cem\u003eClostridia\u003c/em\u003e, including the genera \u003cem\u003eAcetatifactor, Lachnobacterium, Fusimonas, Enterocloster, and Velocimicrobium\u003c/em\u003e (Fig.4e). In contrast, treatment with LC No-Neuro-derived GMEVs was associated with enrichment of a partially distinct set of genera, including \u003cem\u003eTuricibacter\u0026nbsp;\u003c/em\u003eand \u003cem\u003eBifidobacterium\u0026nbsp;\u003c/em\u003e(Fig.4e).\u003c/p\u003e\n\u003cp\u003eTogether, these data indicate that oral exposure to LC-derived GMEVs is sufficient to remodel the faecal microbiota in WT mice, with LC Neuro-derived GMEVs associated with the most distinct compositional shifts.\u003c/p\u003e\n\u003cp\u003eOral administration of GMEVs from individuals with LC induces intestinal inflammation, neurobehavioral alterations and neuroinflammation in vivo\u003c/p\u003e\n\u003cp\u003eWe next tested whether oral exposure to donor-derived GMEVs is sufficient to elicit intestinal and systemic inflammatory changes accompanied by neurobehavioural and neuroinflammatory phenotypes in wild-type mice. Female C57BL/6 mice received oral gavage of GMEVs (5 \u0026micro;g per dose) isolated from LC donors with neurological symptoms (Neuro), LC donors without neurological symptoms (No-Neuro) or pandemic controls (PC), or vehicle alone, 3 times per week for 6 weeks, followed by neurobehavioural testing and tissue collection (Fig.5a).\u003c/p\u003e\n\u003cp\u003eMice receiving LC-derived GMEVs showed evidence of intestinal and systemic inflammatory activation. Colon length was modestly reduced in GMEV-treated groups compared with vehicle, with the largest reduction observed in the Neuro GMEV group (Fig.5b). Circulating markers were also altered as serum lipopolysaccharide-binding protein (LBP) was increased in GMEV-treated groups relative to vehicle, and serum BAFF was significantly elevated in mice receiving Neuro-derived GMEVs compared with controls (Fig.5b). In colonic epithelium, Neuro-derived GMEVs increased expression of inflammasome- and inflammatory-associated genes, including \u003cem\u003eNlrp3\u003c/em\u003e, \u003cem\u003eIl1b\u003c/em\u003e and \u003cem\u003eTnfsf13b\u003c/em\u003e, relative to controls, whereas changes in \u003cem\u003eTnf\u003c/em\u003e and \u003cem\u003eIl6\u0026nbsp;\u003c/em\u003ewere more modest (Fig.5c). Consistent with barrier perturbation, ZO-1 immunostaining showed more frequent junctional discontinuities in mice receiving LC-derived GMEVs, which was most apparent in the Neuro GMEV group, relative to vehicle and PC conditions (Fig.5d).\u003c/p\u003e\n\u003cp\u003eWe next evaluated whether these inflammatory changes were accompanied by alterations in behaviour. Anxiety-like behaviour was evaluated using the open-field test, and short-term spatial memory was assessed using the Y-maze\u003csup\u003e44\u003c/sup\u003e. In the open-field test, mice receiving Neuro-derived GMEVs travelled a greater distance and displayed higher mean velocity than control groups, indicating increased locomotor activity (Fig.5e). Centre-border measures showed more modest differences as time spent in the border zone increased in the Neuro group, whereas centre time and the centre-to-border time ratio showed non-significant trends (Fig.5e), a pattern that can be suggestive of anxiety-like behaviour but is not definitive in isolation. In the Y-maze, Neuro-derived GMEVs increased total alternations but reduced the proportion of successful alternations, with a concurrent increase in indirect revisits, consistent with altered spontaneous alternation behaviour (Fig.5f) and suggestive of impaired short-term spatial working memory. Memory impairment is a common symptom reported by LC Neuro participants in the IPCO cohort (Fig.1b).\u003c/p\u003e\n\u003cp\u003eFinally, building on the behavioural phenotypes and the observation that orally administered bacterial EVs can be detected beyond the gut (Extended Data Fig.9), we assessed glial activation in brain tissue collected after behavioural testing. Mice treated with Neuro-derived GMEVs showed increased GFAP immunoreactivity, with prominent signal in regions proximal to the hippocampus, and increased IBA1-positive microglial signal with an activated morphology in the hindbrain, compared with control groups (Fig.5g-i). These findings are consistent with induction of neuroinflammatory changes following oral exposure to LC-derived vesicles.\u003c/p\u003e\n\u003cp\u003eTogether, these data show that oral administration of GMEVs derived from individuals with LC, most consistently for Neuro-derived preparations, is sufficient to induce intestinal inflammatory responses and barrier perturbation, accompanied by systemic inflammatory signatures, neurobehavioural alterations, and glial activation in vivo. These findings implicate GMEVs as functional mediators linking LC-associated dysbiosis to gut-brain axis perturbations that may contribute to neurological sequelae in LC.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eLC is associated with persistent immune dysregulation, intestinal barrier dysfunction, and neuroinflammation\u003csup\u003e5,8-10,17,20,25\u003c/sup\u003e, yet the mechanisms linking these processes remain incompletely defined. Here we identify GMEVs as functional mediators connecting intestinal dysbiosis to systemic and neuroinflammatory features of LC. Using complementary human, in vitro, and in vivo approaches, we show that GMEVs derived from individuals with LC induce epithelial barrier disruption, immune activation, neurobehavioral alterations, and glial activation, thereby providing mechanistic insight into gut-brain axis perturbations in LC.\u003c/p\u003e\n\n\u003cp\u003eConsistent with previous reports in acute and post-acute syndromes\u003csup\u003e17-20\u003c/sup\u003e, we observed sustained intestinal dysbiosis in individuals with LC that persisted for at least 12 months after infection. Importantly, microbiome alterations correlated with prototypical LC symptoms, particularly neurological manifestations, reinforcing a link between gut microbial composition and disease phenotype. Transfer of LC-associated microbiota into germ-free mice recapitulated features of intestinal barrier dysfunction, systemic inflammation, and neuroinflammation, with effects most pronounced in mice colonization with microbiota from individuals with neurological symptoms. These findings extend prior faecal microbiota transplantation studies\u003csup\u003e45\u003c/sup\u003e by demonstrating symptom-severity-dependent effects and support a contributory role for the gut microbiota in LC pathophysiology. \u003c/p\u003e\n\n\u003cp\u003eEVs represent a major mode of host-microbe communication, yet their role in post-viral syndromes has remained largely unexplored. Althoug EVs derived from host cells have been implicated in acute COVID-19 and LC\u003csup\u003e46,47\u003c/sup\u003e, our data show that microbiota-derived EVs alone are sufficient to induce LC-relevant phenotypes. GMEVs from LC donors promoted inflammasome activation, epithelial permeability, and inflammatory cytokine/chemokine production in intestinal epithelial cells and macrophages, and induced a coordinated inflammatory transcriptional program in human in iPSC-derived microglia. Notably, while GMEVs from LC Neuro and LC No-Neuro donors elicited broadly similar microglial transcriptomic responses, Neuro-derived GMEVs consistently induced stronger epithelial and macrophage activation, suggesting cell-type-specific sensitivity to GMEV-associated inflammatory signals.\u003c/p\u003e\n\n\u003cp\u003eIn vivo, chronic oral administration of LC-derived GMEVs was sufficient to alter gut microbial composition, disrupt intestinal barrier integrity and induce systemic inflammation, recapitulating several features observed in our microbiota transfer model. Fluorescent tracing experiments demonstrated that orally administered bacterial EVs disseminate beyond the gastrointestinal tract and accumulate in peripheral organs, including the brain, consistent with previous reports\u003csup\u003e48\u003c/sup\u003e. Mice receiving GMEVs from LC Neuro donors developed increased locomotor activity, altered exploratory behaviour and impaired spontaneous alternation performance, accompanied by accumulation of GFAP-positive reactive astrocytes and IBA-1positive activated microglia in discrete brain regions. These findings indicate that LC-derived GMEVs induce spatially localized glial activation consistent with neuroinflammatory processes that may not be captured by bulk tissue analyses.\u003c/p\u003e\n\n\u003cp\u003eAcross several experiments, BAFF emerged as a convergent inflammatory mediator. Elevated BAFF levels distinguished LC severity in our cohort and were consistently induced by LC-derived GMEVs in intestinal epithelial cells, macrophages, microglia and in vivo. Excess BAFF is a recognized driver of chronic immune activation, B-cell dysregulation and autoantibody production\u003csup\u003e49-54\u003c/sup\u003e, and has been implicated in other chronic inflammatory conditions, including HIV infection\u003csup\u003e49,50,55\u003c/sup\u003e. Our findings position BAFF at the intersection of intestinal permeability, immune activation and neuroinflammation in LC, and suggest that GMEVs may contribute to BAFF-driven immunopathology. \u003c/p\u003e\n\n\u003cp\u003eMechanistically, GMEVs may engage host cells through multiple pathways, including pattern-recognition receptor signalling triggered by surface-associated microbial components, vesicle internalization, and delivery of biologically active cargo. Differences in EV composition between donor groups, potentially reflecting microbial taxonomic shifts or structural variation in microbial products such as lipopolysaccharide\u003csup\u003e56,57\u003c/sup\u003e, may underlie the graded inflammatory responses observed across GMEVs derived from different LC donor groups. Although LC is undoubtedly multifactorial, our data support a model in which GMEVs amplify and perpetuate inflammatory signalling downstream of intestinal dysbiosis.\u003c/p\u003e\n\n\u003cp\u003eSeveral limitations should be considered. Although the IPCO cohort is longitudinal and clinically characterized, LC is heterogeneous and residual confounding by factors that influence the microbiome, including diet, medication exposures and co-morbidities, cannot be fully excluded. The biodistribution experiment used fluorescently-labelled \u003cem\u003eEscherichia coli\u003c/em\u003e EVs as a surrogate for complex GMEVs. Thus, the kinetics and tissue tropism of human donor-derived vesicles may differ. Neurobehavioural assays provide operational measures of exploration and spontaneous alternation but do not uniquely map to specific neuropsychiatric constructs. Accordingly, the observed open-field and Y-maze changes should be interpreted as behavioural alterations consistent with neuroimmune perturbation rather than definitive measures of anxiety or memory impairment. Finally, the precise microbial sources and vesicle cargo responsible for host activation were not resolved. Defining the molecular determinants of GMEV bioactivity and the host pathways required for their effects will be important for therapeutic translation.\u003c/p\u003e\n\n\u003cp\u003eTogether, these findings identify GMEVs as previously unrecognized effectors linking intestinal dysbiosis to immune dysregulation and neuroinflammation in LC. By integrating human cohort analyses with mechanistic in vitro and in vivo models, this work highlights GMEVs as both biomarkers and potential therapeutic targets. Interventions aimed at restoring microbial homeostasis or modulating downstream inflammatory pathways, including BAFF signalling, may therefore hold promise for mitigating gut\u0026ndash;brain axis dysfunction and neurological symptom burden in long COVID.\u003c/p\u003e\n"},{"header":"Methods","content":"\u003cp\u003eStudy design and population\u003c/p\u003e\n\n\u003cp\u003eIn response to the COVID-19 pandemic, we established the Institut de Recherches Cliniques de Montr\u0026eacute;al\u003cem\u003e \u003c/em\u003e(IRCM) Post-COVID-19 (IPCO) research clinic, which integrates clinical care into a prospective observational cohort study with an associated biobank (IPCO protocol #2021-1092, ClinicalTrials.gov: NCT04736732). Participants are followed longitudinally for up to 24 months with standardized clinical assessments and biospecimen collection. Adults (\u0026sup3;18 years) with confirmed SARS-CoV-2 infection at least 3 months before enrolment and persistent symptoms not attributable to alternative diagnoses were recruited as Long COVID (LC) participants. Pandemic controls (PC) were individuals without persistent symptoms and without reported SARS-CoV-2 infection or positive testing. Participants underwent in-person visits at enrolment and at 6, 12 and 24 months after infection, with a telephone follow-up at 18 months. Baseline demographic and clinical characteristics of participants included in microbiome analyses are provided in Extended Data Table 1.\u003c/p\u003e\n\n\n\u003cp\u003eSymptom definitions and neurological stratification\u003c/p\u003e\n\n\u003cp\u003eParticipants reported symptoms using standardized questionnaires administered at each visit. For analyses stratifying LC participants by neurological symptom burden, 8 self-reported symptoms were considered: post-exertional malaise, trouble with concentration, trouble with memory, trouble with sleep, anxiety, brain fog, confusion and depression. Participants reporting \u0026sup3;1 symptom were classified as LC Neuro, whereas those reporting non were classified LC No-Neuro. For symptom-burden analyses, LC participants reporting symptoms were categorized as Mild Neuro (\u0026lt;3 symptoms) or Severe Neuro (\u0026ge;3 symptoms). \u003c/p\u003e\n\n\n\u003cp\u003eSample collection and processing\u003c/p\u003e\n\n\u003cp\u003eAt each study visit, participants provided stool, blood, saliva, and urine samples. Serum and plasma were isolated by centrifugation. Peripheral blood mononuclear cells (PBMCs) were isolated using SepMate tubes (STEMCELL Technologies). All biospecimens were aliquoted and stored at -80\u0026deg;C, and PBMCs were cryopreserved in liquid nitrogen. \u003c/p\u003e\n\n\n\u003cp\u003eStool DNA extraction and microbiome sequencing\u003c/p\u003e\n\n\u003cp\u003eHuman samples\u003c/p\u003e\n\n\u003cp\u003eStool DNA was extracted using the QIAamp PowerFecal Pro DNA Kit (QIAGEN) with mechanical homogenization. Shotgun metagenomic libraries were prepared after quality control and sequenced on an Illumina NovaSeq 6000 platform (CosmosID). Host reads were removed by alignment to the human reference genome (GRCh38). Read quality was assessed using FastQC\u003csup\u003e58\u003c/sup\u003e. Functional profiling was performed using HUMAnM (v3.9)\u003csup\u003e59\u003c/sup\u003e and taxonomic profiling using MetaPhlAn (v4.1.1)\u003csup\u003e60\u003c/sup\u003e. \u003c/p\u003e\n\n\u003cp\u003eMicrobiome analyses were performed in R (v4.2) using phyloseq (v1.48.0)\u003csup\u003e61\u003c/sup\u003e and tidyverse packages\u003csup\u003e62\u003c/sup\u003e. Alpha diversity was assessed using the Simpson index Pielou\u0026rsquo;s evenness, and beta diversity using Bray-Curtis dissimilarity with principal coordinate analysis (PCoA). Differentially abundant taxa were identified using Linear Discriminant Analysis (LDA) Effect Size (LEfSe)-based approaches implemented in edgeR\u003csup\u003e63\u003c/sup\u003e and microbiomeMarker\u003csup\u003e64\u003c/sup\u003e, with a Kruskal-Wallis significance threshold of \u003cem\u003eP\u003c/em\u003e \u0026le; 0.05 and an LDA score cutoff as indicated in the corresponding figure legends.\u003c/p\u003e\n\n\n\u003cp\u003eMouse stool 16S rRNA sequencing\u003c/p\u003e\n\n\u003cp\u003eMouse faecal pellets were collected into sterile tubes, snap-frozen on dry ice and stored at -80\u0026deg;C. DNA was extracted using the DNeasy PowerSoil Pro QIAcube HT Kit (QIAGEN). The V4 region of the 16S rRNA gene was amplified and sequenced on an Illumina MiSeq platform (McGill Centre for Microbiome Research). Read quality was assessed using FastQC\u003csup\u003e58\u003c/sup\u003e, adapters and primers were trimmed with Trimmomatic\u003csup\u003e65\u003c/sup\u003e. Paired-end reads were merged with PANDAseq\u003csup\u003e66\u003c/sup\u003e and chimeras removed as described previously\u003csup\u003e67\u003c/sup\u003e. Sequences were clustered using CD-HIT-EST\u003csup\u003e68\u003c/sup\u003e and taxonomically classified using Kraken2\u003csup\u003e69\u003c/sup\u003e against an in-house RefSeq database (updated November 2023). \u003c/p\u003e\n\n\u003cp\u003eAlpha diversity was assessed using Shannon diversity and Pielou\u0026rsquo;s evenness. Beta diversity was evaluated using Bray-Curtis dissimilarity and visualized by PCoA. Group differences were assessed by permutational multivariate analysis of variance (PERMANOVA; adonis2, vegan v2.6-6.1; 1000 permutations)\u003csup\u003e70\u003c/sup\u003e, with pairwise comparisons adjusted for multiple testing as indicated. Differentially abundant taxa were identified using LEfSe (edgeR\u003csup\u003e63\u003c/sup\u003e and microbiomeMarker\u003csup\u003e64\u003c/sup\u003e; LDA cutoff and significance thresholds as indicated in the corresponding figure legends).\u003c/p\u003e\n\n\u003cp\u003eAnimals and housing\u003c/p\u003e\n\n\u003cp\u003eWild-type (WT) C57BL/6 female mice (6-8 weeks old) were obtained from The Jackson Laboratory and housed under specific pathogen-free conditions at the IRCM animal facility with a 12-hour light/dark cycle with controlled temperature and humidity and ad libitum access to food and water. Germ-free (GF) C57BL/6 female mice (5-7 weeks old) were obtained from the Germ-Free and Gnotobiotic Platform at the University of Calgary (K. McCoy) and maintained in flexible film isolators (Class Biologically Clean) under sterile conditions at the IRCM gnotobiotic facility. Mice were routinely screened for contamination. All animal procedures were approved by the IRCM Animal Care Committee.\u003c/p\u003e\n\n\n\u003cp\u003eFaecal microbiota transplantation into germ-free mice\u003c/p\u003e\n\n\u003cp\u003eMouse experimental group sizes and anonymised donor sample identifiers are summarized in Extended Data Table 2. GF mice were orally gavaged 3 times per week with 200\u0026mu;l of a stool suspension prepared by resuspending 200mg of human donor stool in 1mL sterile PBS. After the first gavage, mice were transferred to positive-pressure cages to prevent cross-contamination. Microbiota were allowed to engraft for 3 weeks before mice were transferred to irradiated racks for acclimation, behavioural testing and tissue collection.\u003c/p\u003e\n\n\n\u003cp\u003eGMEV isolation and characterization\u003c/p\u003e\n\n\u003cp\u003eGut microbiota-derived extracellular vesicles (GMEVs) were isolated from human stool by size-exclusion chromatography. Stool suspensions were sequentially centrifuged and filtered (0.45\u0026mu;m and 0.22\u0026mu;m) to remove bacteria and debris, concentrated by ultrafiltration (100kDa cut-off), and fractionated using qEV Original 35nm Gen 2 column (IZON). GMEV-enriched fractions were pooled, aliquoted, and stored at -80\u0026deg;C. Particle size and concentration were assessed by nanoparticle tracking analysis (ZetaView PMX120). Vesicular morphology was confirmed by transmission electron microscopy (Tecnai G2 Spirit Twin). Protein concentration was measured using the Pierce Bovine Serum Albumin Protein assay (Thermo Fisher Scientific).\u003c/p\u003e\n\n\n\u003cp\u003eNeurobehavioural testing\u003c/p\u003e\n\n\u003cp\u003eNeurobehavioural testing was performed after microbiota engraftment (GF colonization experiments) or after the oral gavage regimen (WT GMEV experiments), as indicated in the relevant figure legends. Testing was recorded using the EthoVision and analysed blinded to group allocation where feasible.\u003c/p\u003e\n\n\n\u003cp\u003eY-maze spontaneous alternation \u003c/p\u003e\n\n\u003cp\u003eThe Y-maze test was used to assess spontaneous alternation behaviour, which is sensitive to short-term spatial working memory\u003csup\u003e44\u003c/sup\u003e. Mice were placed in the centre of a 3-armed maze and allowed to explore freely for 8 minutes. Arm entries were scored by EthoVision. Consecutive triplets of arm entries were defined as alternations; triplets comprising 3 different arms were scored as successful alternations. Performance was calculated as the ratio of successful alternations to total alternations, with direct and indirect revisits also quantified. A lux meter was used to ensure that illumination was standardized across all 3 arms (approximately 150 lux).\u003c/p\u003e\n\n\n\u003cp\u003eOpen-field test\u003c/p\u003e\n\n\u003cp\u003eExploration and anxiety-like behaviour were assessed using the open-field test\u003csup\u003e71\u003c/sup\u003e. Mice were placed in the arena and recorded by EthoVision. Time spent in the centre versus periphery (border), the number of centre-border transitions, total distance travelled, and velocity were quantified. A lux meter was used to ensure that illumination was standardized across the whole arena (approximately 150 lux).\u003c/p\u003e\n\n\n\u003cp\u003eOral gavage of GMEVs in wild-type mice\u003c/p\u003e\n\n\u003cp\u003eGroup sizes and anonymised donor sample identifiers are provided in Extended Data Table 2. Female WT C56BL/6 mice (6-8 weeks old; The Jackson Laboratory), maintained under specific pathogen-free conditions, received GMEVs by oral gavage (5\u0026mu;g per dose in 200\u0026mu;l sterile PBS) or vehicle (sterile PBS) 3 times per week for 6 weeks. Following the treatment period, mice underwent behavioural testing prior to euthanasia and tissue collection. Colon length was measured at necropsy. Colon and brain tissue were processed for histology and molecular analyses as described below.\u003c/p\u003e\n\n\u003cp\u003eBiodistribution of orally administered EVs\u003c/p\u003e\n\n\u003cp\u003eTo assess biodistribution following oral delivery, extracellular vesicles were isolated from Escherichia coli (ATCC 25922) culture supernatants by ultrafiltration and size-exclusion chromatography (qEV 35\u0026thinsp;nm Gen 2; Izon). Vesicles were labelled with Vybrant DiD (Invitrogen) according to the manufacturer\u0026rsquo;s instructions and administered to female BALB/c mice by oral gavage (10\u0026micro;g dose of EVs). Fluorescence was acquired 8h after gavage using a Xenogen IVIS 200 system (PerkinElmer) under isoflurane anaesthesia. Mice were then euthanized, and organs were harvested for ex vivo fluorescence imaging.\u003c/p\u003e\n\n\n\u003cp\u003eBlood collection and serum isolation\u003c/p\u003e\n\n\u003cp\u003eMice were euthanized by terminal cardiac puncture. Whole blood was allowed to clot at room temperature for 30 minutes, incubated at 37\u0026deg;C for 10 minutes, and then incubated at 4\u0026deg;C for 15 minutes before centrifugation (3,000g, 20 minutes). Serum was stored at -80\u0026deg;C until analysis.\u003c/p\u003e\n\n\n\u003cp\u003eELISA measurements of soluble BAFF and LBP \u003c/p\u003e\n\n\u003cp\u003eSerum B-cell activation factor (BAFF) and lipopolysaccharide-binding protein (LBP) were quantified using the Quantikine Mouse BAFF/BLyS/TNFSF13B ELISA kit (R\u0026amp;D Systems) and the Mouse LBP ELISA kit (Abcam), respectively, following the manufacturers\u0026rsquo; instructions.\u003c/p\u003e\n\n\n\u003cp\u003eColon epithelial cell isolation\u003c/p\u003e\n\n\u003cp\u003eColon epithelial cells were isolated using a dithiothreitol (DTT)/EDTA-based epithelial stripping approach. Colons were washed in HBSS without Ca\u003csup\u003e++\u003c/sup\u003e and Mg\u003csup\u003e++\u003c/sup\u003e supplemented with EDTA (2mM) and HEPES (25mM). Tissue was incubated at 37\u0026deg;C for 15 minutes in stripping buffer (HBSS without Ca\u003csup\u003e++\u003c/sup\u003e/Mg\u003csup\u003e++\u003c/sup\u003e supplemented with HEPES (15mM), EDTA (5mM) and DTT (1mM)) and vortexed to release epithelial cells. The epithelial fraction was filtered (100\u0026mu;m), washed in PBS, and lysed in RLT buffer (QIAGEN) for RNA extraction.\u003c/p\u003e\n\n\n\u003cp\u003eTissue embedding, immunofluorescence staining and imaging\u003c/p\u003e\n\n\u003cp\u003eColon tissue and brain hemisphere were embedded in OCT embedding medium (Scigen Scientific) and stored at -80\u0026deg;C. Colon cryosections (10\u0026micro;m) were prepared at the IRCM Histology Core Facility, fixed in cold acetone (-20\u0026deg;C), air dried, and stored at -80\u0026deg;C. Cryosections (14\u0026mu;m) of brain hemispheres were prepared at the Institut de Recherche en Immunologie et en Canc\u0026eacute;rologie (IRIC) Histology Core Facility and processed identically. \u003c/p\u003e\n\n\u003cp\u003eImmunofluorescence staining was performed at the Centre de Recherche du Centre Hospitalier de l\u0026rsquo;Universit\u0026eacute; de Montr\u0026eacute;al (CRCHUM) Molecular Pathology Core Facility using a Discovery Ultra automated stainer (Ventana/Roche). Sections were blocked in PBS containing 1% bovine serum albumin (BSA) for 30 minutes at room temperature, incubated with primary antibodies followed by species-appropriate secondary antibodies for 2 hours each at room temperature. Nuclei were counterstained with DAPI (1:3,000) for 10 minutes. Sudan Black (0.1% in 70% ethanol) was applied to reduce autofluorescence, and sections were mounted with Fluoromount (Sigma). \u003c/p\u003e\n\n\u003cp\u003eSlides were scanned using an Aperio Verso 200 scanner (Leica Biosystems) with a 20\u0026acute;/0.8 NA objective at a resolution of 0.275\u0026mu;m per pixel. Image visualization and analysis were performed blinded to group allocation using Aperio ImageScope software (Leica Biosystems). \u003c/p\u003e\n\n\u003cp\u003ePrimary antibodies used included rabbit anti-mouse zonula occludens-1 (ZO-1) and ionized calcium-binding adaptor molecule 1 (IBA1) and an anti-glial fibrillary acidic protein (GFAP) monoclonal antibody; Alexa Fluor-conjugated goat anti-rabbit secondary antibodies were used as indicated. Antibody details are provided in Extended Data Table 4b.\u003c/p\u003e\n\n\n\n\u003cp\u003eCell culture and stimulation with GMEVs\u003c/p\u003e\n\u003cp\u003eTHP-1-derived human macrophages\u003c/p\u003e\n\n\u003cp\u003eTHP-1 cells were maintained in complete RPMI 1640 supplemented with 10% heat-inactivated foetal bovine serum (FBS), 1% penicillin-streptomycin and 14.3\u0026mu;M 2-mercaptoethanol. For differentiation, THP-1 cells were seeded at 5 x 10\u003csup\u003e5\u003c/sup\u003e cells per well (12-well plates) and treated with phorbol 12-myristate 13-acetate (PMA; 100nM) for 48 hours, washed and rested in fresh complete medium before stimulation. Macrophages were stimulated with GMEVs (1\u0026mu;g/mL) for 5h (RNA) or 16h (secreted proteins). For intracellular cytokine staining, cells were treated with BD GolgiPlug protein transport inhibitor (containing brefeldin A) during the final 5h of stimulation.\u003c/p\u003e\n\n\n\u003cp\u003eHuman intestinal organoids and epithelial monolayers\u003c/p\u003e\n\n\u003cp\u003eHuman intestinal organoids (HIOs) were generated from human induced pluripotent stem cells (iPSCs) derived from fibroblasts from a healthy female donor (generously provided by H. Malech, NIH) using the STEMdiff\u0026trade; Intestinal Organoid Kit (STEMCELL Technologies) following the manufacturer\u0026rsquo;s instructions. iPSCs were maintained on Matrigel in mTeSR1 medium and differentiated into intestinal lineage using the STEMdiff\u0026trade; Intestinal Organoid Kit following the manufacturer\u0026rsquo;s protocol, including definitive endoderm induction and subsequent mid-hindgut patterning. Mid-hindgut spheroids were embedded in Matrigel domes and matured in Intestinal Organoid Growth Medium, with medium changes every 2-3 days and weekly passaging. For stimulation experiments, HIOs were dissociated into epithelial monolayers, plated onto Matrigel-coated 24-well plates and treated with GMEVs (1\u0026thinsp;\u0026micro;g/mL) for 5h (RNA) or 16h (secreted proteins).\u003c/p\u003e\n\n\n\u003cp\u003eTransepithelial electrical resistance \u003c/p\u003e\n\n\u003cp\u003eCaco-2 cells were seeded at 1.5x10\u003csup\u003e5\u003c/sup\u003e cells/cm\u003csup\u003e2\u003c/sup\u003e on Transwell inserts (0.4\u0026mu;m pore; 12-well format; Sarstedt) and maintained for \u0026ge;3 weeks until transepithelial electrical resistance (TEER) stabilized. Cells were cultured in Eagle\u0026rsquo;s Modified Essential Medium (EMEM) supplemented with 20% heat-inactivated FBS and penicillin-streptomycin. Baseline TEER was recorded before treatment. Cells were treated with GMEVs (3\u0026thinsp;\u0026micro;g/mL), and TEER measured 24h later using an EVOM2 voltohmmeter with STX2 electrodes. Values reflect the mean of 3 measurements per insert.\u003c/p\u003e\n\n\n\u003cp\u003eRNA Extraction and quantitative PCR\u003c/p\u003e\n\n\u003cp\u003eTotal RNA was extracted from colon epithelial cells, HIOs, iPSC-derived microglia, and THP-1-derived macrophages using the RNeasy Plus Mini Kit (QIAGEN) according to the manufacturer\u0026apos;s instructions. RNA concentration and purity were assessed using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific). Complementary DNA was synthesized from 1\u0026mu;g total RNA using the iScript Reverse Transcription SuperMix for RT-qPCR (BioRad). Quantitative PCR was performed using SsoAdvanced Universal SYBR Green Supermix (BioRad) on a StepOnePlus Real-Time PCR System (Applied Biosystems). Relative gene expression was calculated using the 2\u003csup\u003e(-\u0026Delta;\u0026Delta;Ct)\u003c/sup\u003e method with GAPDH as the reference gene. Primer sequences are listed in Extended Data Table 3.\u003c/p\u003e\n\n\u003cp\u003e\u003cem\u003e \u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eDifferentiation of induced pluripotent stem cell-derived microglia\u003c/p\u003e\n\n\u003cp\u003eHuman iPSCs were generated from a healthy female donor and maintained using standard procedures as described\u003csup\u003e72\u003c/sup\u003e. Microglia were differentiated from iPSCs using a two-step protocol adapted from established methods\u003csup\u003e73\u003c/sup\u003e. Briefly, iPSCs were differentiated into iPSC-derived hematopoietic progenitor cells (iHPCs) using the STEMdiff Hematopoietic Kit (STEMCELL Technologies), with minor modifications. On day -1, iPSCs were dissociated using Gentle Cell Dissociation Reagent (STEMCELL Technologies) and seeded onto Matrigel-coated 6-well plates in mTeSR\u0026trade; Plus or Essential 8 medium supplemented with Y-27632 (10\u0026mu;M, Selleckchem) at a density designed to yield colonies of fewer than 100 cells per cm\u0026sup2; by day 0. On day 0, medium (1mL) was replaced with STEMdiff Hematopoietic Medium A (2mL per well). On day 2, half of the medium (1mL) was replaced with fresh Medium A. On day 3, cultures were switched entirely to STEMdiff Hematopoietic Medium B (2mL per well). On days 5 and 7, half of the supernatant (1mL) was replaced with fresh Medium B. On day 9, an additional 1mL of fresh Medium B was added. On day 10, non-adherent iHPCs present in the supernatant were collected, centrifuged at 300g for 5 minutes, and either cryopreserved (Bambanker, Fujifilm Wako Chemicals) or used for microglial differentiation. This harvesting procedure was repeated on days 12 and 14. For microglial differentiation, iHPCs were resuspended in microglia differentiation medium (as previously described) at a density of 5x10\u003csup\u003e4\u003c/sup\u003e cells/mL and plated onto Matrigel-coated 6-well plates (2mL per well), defining day 0 of differentiation. Cultures were supplemented every other day with 1mL of fresh differentiation medium from day 0 to day 10. On day 12, supernatants were collected, cells were pelleted by centrifugation (300g, 5 minutes), resuspended in fresh differentiation medium, and returned to culture. This procedure was repeated on day 24. Cells were considered mature by day 28 and were maintained with media changes every other day until use. All cultures were maintained at 37\u0026deg;C in a humidified atmosphere containing 5% CO\u003csub\u003e2\u003c/sub\u003e. For downstream experiments, cells were detached using PBS containing 2mM EDTA and replated at the desired density. \u003c/p\u003e\n\n\n\u003cp\u003eBulk RNA sequencing of GMEV-treated iMGLs\u003c/p\u003e\n\n\u003cp\u003eInduced pluripotent stem cell-derived microglia (iMGLs) were treated with GMEVs (1\u0026mu;g/mL) from LC donors (n=10) or PC donors (n=4) for 16h. RNA (\u0026pound;100ng) was submitted for ribodepletion library preparation and sequencing (target ~50 million reads per sample; IRCM Molecular Biology platform). Read quality was assessed using FastQC (v0.12.1). Reads were aligned to GRCh38 using STAR (v2.7.11b), gene counts were generated using featureCounts v2.0.6; GRCh38 release 110), and differential expression was performed using the DESeq2. Heatmaps were generated using z-scored normalized counts. Functional enrichment analysis was performed using gprofiler2. \u003c/p\u003e\n\n\n\u003cp\u003eCytokine and chemokine quantification in cell culture supernatants\u003c/p\u003e\n\n\u003cp\u003eSecreted cytokines/chemokines were quantified using MSD U-PLEX panels. For THP-1 macrophages and HIOs, analytes included granulocyte-macrophage colony stimulating factor (GM-CSF), granulocyte-colony stimulating factor (G-CSF), interferon gamma (IFN\u0026gamma;), IL-10, IL-1\u0026beta;, IL-6, IL-8, CXCL10, CCL2, and tumor necrosis factor (TNF). For iMGL supernatants, the panel included B-cell activation factor (BAFF), IL-1\u0026beta;, IL-6, IL-8, TNF, s100A12, matrix metalloproteinase 9 (MMP9), CCL2 and CXCL1, with R-PLEX detection of C1q. Plates were read on a MESO QuickPlex SQ 120 and analysed using MSD Discovery Workbench 4.0.\u003c/p\u003e\n\n\n\u003cp\u003eFlow cytometry of THP-1-derived macrophages\u003c/p\u003e\n\n\u003cp\u003eCells were stained with LIVE/DEAD Fixable Aqua (Thermo Fisher Scientific), blocked with human Fc block (BD Biosciences) with 20% heat inactivated FBS and 50\u0026mu;g mouse and/or rat IgG, and stained for surface BAFF, followed by fixation/permeabilization (Cytofix/Cytoperm, BD Biosciences) and intracellular staining for IL-1ꞵ, IL-6 and TNF. Data were acquired on a BD LSRFortessa and analysed in FlowJo (v10.8.1). Antibody details are provided in Extended Data Table 4a.\u003c/p\u003e\n\n\n\u003cp\u003eStatistical analyses \u003c/p\u003e\n\n\u003cp\u003eGroup comparisons were performed using one-way ANOVA with Tukey\u0026rsquo;s post hoc test for approximately normally distributed data, or Kruskal\u0026ndash;Wallis tests with Dunn\u0026rsquo;s post hoc correction otherwise. For two-group comparisons, Wilcoxon rank-sum (two-sided) tests were used unless stated otherwise. Correlations were assessed using Pearson\u0026rsquo;s or Spearman\u0026rsquo;s tests, as appropriate. Categorical variables were analysed using Fisher\u0026rsquo;s exact test. Unless otherwise indicated, data are shown as medians with interquartile ranges. All tests were two-sided and \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05 was considered statistically significant. Analyses were performed using GraphPad Prism v10.2.0.\u003c/p\u003e\n"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eE.L.F is supported by a Tier 2 Canada Research Chair in Role of the Microbiome in Inborn Errors of Immunity and Post-Infectious Conditions, the Canadian Institutes of Health Research (CIHR), the \u003cem\u003eFonds de Recherche du Qu\u0026eacute;bec (FRQ)\u003c/em\u003e. K.D.L. is supported by the IRCM Foundation. M.A. is supported by CIHR. A.D. is supported by\u003cem\u003e\u0026nbsp;\u003c/em\u003ethe \u003cem\u003eFonds de Recherche du Qu\u0026eacute;bec (FRQ)\u003c/em\u003e. I.B. is supported by the IRCM Foundation. The work was supported by CIHR (PJT-191724), the Canada Research Chairs Program, the FRQ Clinical Research Scholars - Junior 1 Establishment Funds for Young Investigators, the John R. Evans Leaders Fund from the Canadian Foundation for Innovation (CFI), the J-Louis L\u0026eacute;vesque Foundation Research Chair, the Mirella and Lino Saputo Foundation, and the IRCM Foundation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe thank members of the IRCM animal facility (Mariane Canuel, Eve-Marie Charbonneau, Manon Laprise, and Jo-Anny Bisson), the IRCM Flow Cytometry Platform (\u0026Eacute;ric Massicotte and Julie Lord), and the McGill Genome Centre, Centre for Microbiome Research, for technical support. We are grateful to Kathy McCoy for providing germ-free mice. We thank Christian Charbonneau, Dr. Marianne Isaac and Melina Narlis of the IRIC Microscopy and Histology Core Facilities for guidance and for performing sagittal brain cryosections and H\u0026amp;E staining. We also thank V\u0026eacute;ronique Barr\u0026egrave;s and Liliane Meunier of the CRCHUM Molecular Pathology Core Facility for immunolabeling, slide scanning, and assistance with paraffin and OCT processing of colon and brain, and Anabelle Bouchard-Bourque of the IRCM Histology Core Facility for additional paraffin and OCT sectioning. We acknowledge Dominic Filion and Mattew Duguay of the IRCM Imaging Platform for imaging support. We thank Dr. Mieczyslaw Marcinkiewicz, and Dennis A. Drewnik for advice and assistance with brain immunofluorescence interpretation, fixation and embedding. We also thank the Centre for Applied Nanomedicine at the Research Institute of the McGill University Hospital Centre for assistance with nanoparticle tracking analysis, and Dr. S. Kelly Sears and Dr. Jeannie Mui from the Facility for Electron Microscopy Research at McGill University for their assistance with electron microscopy.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eM.A. and E.L.F. conceived the study and designed the research. M.A., K.D-L., I.B., A.V., A.D., E.D, L.P., V.E.P., F.B., J.S., and A.A. performed experiments. M.A., K.D-L., I.B., P.C., J.P. and E.L.F. analysed the data. M.A., P.C., and E.L.F. wrote the initial manuscript draft, and all authors contributed to manuscript revision. E.L.F. secured funding for the study. E.L.F. supervised the research.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCompeting interests\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interest.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eData availability\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll raw sequencing data has been deposited in SRA under BioProject PRJNA1236664 (submissions: SUB15113758 and SUB15164815) and will be released upon publication. Any additional data supporting the findings of this study are available from the corresponding author upon reasonable request.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCode availability\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eCustom code used for microbiota processing, statistical analyses, and figure generation will be deposited on GitHub and made publicly available upon publication. Scripts are available from the corresponding author upon reasonable request prior to release.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eWorld Health, O. A clinical case definition of post COVID-19 condition by a Delphi consensus. (World Health Organization, Geneva, 2021).\u003c/li\u003e\n\u003cli\u003eEly, E. W., Brown, L. M., Fineberg, H. V., National Academies of Sciences, E. \u0026amp; Medicine Committee on Examining the Working Definition for Long, C. Long Covid Defined. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cstrong\u003e391\u003c/strong\u003e, 1746-1753 (2024). https://doi.org/10.1056/NEJMsb2408466\u003c/li\u003e\n\u003cli\u003eTaquet, M.\u003cem\u003e et al.\u003c/em\u003e Neurological and psychiatric risk trajectories after SARS-CoV-2 infection: an analysis of 2-year retrospective cohort studies including 1\u0026amp;#x2008;284\u0026amp;#x2008;437 patients. \u003cem\u003eThe Lancet Psychiatry\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 815-827 (2022). https://doi.org/10.1016/S2215-0366(22)00260-7\u003c/li\u003e\n\u003cli\u003eDavis, H. E., McCorkell, L., Vogel, J. M. \u0026amp; Topol, E. J. Long COVID: major findings, mechanisms and recommendations. \u003cem\u003eNat Rev Microbiol\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 133-146 (2023). https://doi.org/10.1038/s41579-022-00846-2\u003c/li\u003e\n\u003cli\u003eKuang, S. E., S.; Clarke, J.; Zakaria, D.; Demers, A.; Aziz, S. Experiences of Canadians with long-term symptoms following COVID-19. \u003cem\u003eInsights on Canadian Society\u003c/em\u003e (2023).\u003c/li\u003e\n\u003cli\u003eCOVID-19, W. H. O. W. c. c. d. w. g. o. p. \u0026amp; condition. (2021).\u003c/li\u003e\n\u003cli\u003eDavis, H. E.\u003cem\u003e et al.\u003c/em\u003e Characterizing long COVID in an international cohort: 7 months of symptoms and their impact. \u003cem\u003eEClinicalMedicine\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 101019 (2021). https://doi.org/10.1016/j.eclinm.2021.101019\u003c/li\u003e\n\u003cli\u003eLopez-Leon, S.\u003cem\u003e et al.\u003c/em\u003e More than 50 long-term effects of COVID-19: a systematic review and meta-analysis. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 16144 (2021). https://doi.org/10.1038/s41598-021-95565-8\u003c/li\u003e\n\u003cli\u003eAl-Aly, Z., Xie, Y. \u0026amp; Bowe, B. High-dimensional characterization of post-acute sequelae of COVID-19. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e594\u003c/strong\u003e, 259-264 (2021). https://doi.org/10.1038/s41586-021-03553-9\u003c/li\u003e\n\u003cli\u003eAl-Aly, Z. \u0026amp; Topol, E. Solving the puzzle of Long Covid. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e383\u003c/strong\u003e, 830-832 (2024). https://doi.org/10.1126/science.adl0867\u003c/li\u003e\n\u003cli\u003eCai, M., Xie, Y., Topol, E. J. \u0026amp; Al-Aly, Z. Three-year outcomes of post-acute sequelae of COVID-19. \u003cem\u003eNat Med\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 1564-1573 (2024). https://doi.org/10.1038/s41591-024-02987-8\u003c/li\u003e\n\u003cli\u003eKlein, J.\u003cem\u003e et al.\u003c/em\u003e Distinguishing features of long COVID identified through immune profiling. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e623\u003c/strong\u003e, 139-148 (2023). https://doi.org/10.1038/s41586-023-06651-y\u003c/li\u003e\n\u003cli\u003eYin, K.\u003cem\u003e et al.\u003c/em\u003e Long COVID manifests with T cell dysregulation, inflammation and an uncoordinated adaptive immune response to SARS-CoV-2. \u003cem\u003eNat Immunol\u003c/em\u003e \u003cstrong\u003e25\u003c/strong\u003e, 218-225 (2024). https://doi.org/10.1038/s41590-023-01724-6\u003c/li\u003e\n\u003cli\u003ePhetsouphanh, C.\u003cem\u003e et al.\u003c/em\u003e Immunological dysfunction persists for 8 months following initial mild-to-moderate SARS-CoV-2 infection. \u003cem\u003eNat Immunol\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 210-216 (2022). https://doi.org/10.1038/s41590-021-01113-x\u003c/li\u003e\n\u003cli\u003eWoodruff, M. C.\u003cem\u003e et al.\u003c/em\u003e Chronic inflammation, neutrophil activity, and autoreactivity splits long COVID. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 4201 (2023). https://doi.org/10.1038/s41467-023-40012-7\u003c/li\u003e\n\u003cli\u003eTalla, A.\u003cem\u003e et al.\u003c/em\u003e Persistent serum protein signatures define an inflammatory subcategory of long COVID. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 3417 (2023). https://doi.org/10.1038/s41467-023-38682-4\u003c/li\u003e\n\u003cli\u003ePretorius, E.\u003cem\u003e et al.\u003c/em\u003e Persistent clotting protein pathology in Long COVID/Post-Acute Sequelae of COVID-19 (PASC) is accompanied by increased levels of antiplasmin. \u003cem\u003eCardiovasc Diabetol\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 172 (2021). https://doi.org/10.1186/s12933-021-01359-7\u003c/li\u003e\n\u003cli\u003eBraga, J.\u003cem\u003e et al.\u003c/em\u003e Neuroinflammation After COVID-19 With Persistent Depressive and Cognitive Symptoms. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e \u003cstrong\u003e80\u003c/strong\u003e, 787-795 (2023). https://doi.org/10.1001/jamapsychiatry.2023.1321\u003c/li\u003e\n\u003cli\u003eSoung, A. L.\u003cem\u003e et al.\u003c/em\u003e COVID-19 induces CNS cytokine expression and loss of hippocampal neurogenesis. \u003cem\u003eBrain\u003c/em\u003e \u003cstrong\u003e145\u003c/strong\u003e, 4193-4201 (2022). https://doi.org/10.1093/brain/awac270\u003c/li\u003e\n\u003cli\u003eChang, R.\u003cem\u003e et al.\u003c/em\u003e Risk of autoimmune diseases in patients with COVID-19: A retrospective cohort study. \u003cem\u003eEClinicalMedicine\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, 101783 (2023). https://doi.org/10.1016/j.eclinm.2022.101783\u003c/li\u003e\n\u003cli\u003eSon, K.\u003cem\u003e et al.\u003c/em\u003e Circulating anti-nuclear autoantibodies in COVID-19 survivors predict long COVID symptoms. \u003cem\u003eEur Respir J\u003c/em\u003e \u003cstrong\u003e61\u003c/strong\u003e (2023). https://doi.org/10.1183/13993003.00970-2022\u003c/li\u003e\n\u003cli\u003eNayyerabadi, M.\u003cem\u003e et al.\u003c/em\u003e Vaccination after developing long COVID: Impact on clinical presentation, viral persistence, and immune responses. \u003cem\u003eInt J Infect Dis\u003c/em\u003e \u003cstrong\u003e136\u003c/strong\u003e, 136-145 (2023). https://doi.org/10.1016/j.ijid.2023.09.006\u003c/li\u003e\n\u003cli\u003eSwank, Z.\u003cem\u003e et al.\u003c/em\u003e Persistent Circulating Severe Acute Respiratory Syndrome Coronavirus 2 Spike Is Associated With Post-acute Coronavirus Disease 2019 Sequelae. \u003cem\u003eClin Infect Dis\u003c/em\u003e \u003cstrong\u003e76\u003c/strong\u003e, e487-e490 (2023). https://doi.org/10.1093/cid/ciac722\u003c/li\u003e\n\u003cli\u003eNatarajan, A.\u003cem\u003e et al.\u003c/em\u003e Gastrointestinal symptoms and fecal shedding of SARS-CoV-2 RNA suggest prolonged gastrointestinal infection. \u003cem\u003eMed\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 371-387 e379 (2022). https://doi.org/10.1016/j.medj.2022.04.001\u003c/li\u003e\n\u003cli\u003eStein, S. R.\u003cem\u003e et al.\u003c/em\u003e SARS-CoV-2 infection and persistence in the human body and brain at autopsy. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e612\u003c/strong\u003e, 758-763 (2022). https://doi.org/10.1038/s41586-022-05542-y\u003c/li\u003e\n\u003cli\u003eZollner, A.\u003cem\u003e et al.\u003c/em\u003e Postacute COVID-19 is Characterized by Gut Viral Antigen Persistence in Inflammatory Bowel Diseases. \u003cem\u003eGastroenterology\u003c/em\u003e \u003cstrong\u003e163\u003c/strong\u003e, 495-506 e498 (2022). https://doi.org/10.1053/j.gastro.2022.04.037\u003c/li\u003e\n\u003cli\u003eYeoh, Y. K.\u003cem\u003e et al.\u003c/em\u003e Gut microbiota composition reflects disease severity and dysfunctional immune responses in patients with COVID-19. \u003cem\u003eGut\u003c/em\u003e \u003cstrong\u003e70\u003c/strong\u003e, 698-706 (2021). https://doi.org/10.1136/gutjnl-2020-323020\u003c/li\u003e\n\u003cli\u003eLiu, Q.\u003cem\u003e et al.\u003c/em\u003e Gut microbiota dynamics in a prospective cohort of patients with post-acute COVID-19 syndrome. \u003cem\u003eGut\u003c/em\u003e \u003cstrong\u003e71\u003c/strong\u003e, 544-552 (2022). https://doi.org/10.1136/gutjnl-2021-325989\u003c/li\u003e\n\u003cli\u003eHonda, K. \u0026amp; Littman, D. R. The microbiota in adaptive immune homeostasis and disease. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e535\u003c/strong\u003e, 75-84 (2016). https://doi.org/10.1038/nature18848\u003c/li\u003e\n\u003cli\u003eMalard, F., Dore, J., Gaugler, B. \u0026amp; Mohty, M. Introduction to host microbiome symbiosis in health and disease. \u003cem\u003eMucosal Immunol\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 547-554 (2021). https://doi.org/10.1038/s41385-020-00365-4\u003c/li\u003e\n\u003cli\u003eDi Vincenzo, F., Del Gaudio, A., Petito, V., Lopetuso, L. R. \u0026amp; Scaldaferri, F. Gut microbiota, intestinal permeability, and systemic inflammation: a narrative review. \u003cem\u003eInternal and Emergency Medicine\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 275-293 (2024). https://doi.org/10.1007/s11739-023-03374-w\u003c/li\u003e\n\u003cli\u003eGiron, L. B.\u003cem\u003e et al.\u003c/em\u003e Plasma Markers of Disrupted Gut Permeability in Severe COVID-19 Patients. \u003cem\u003eFront Immunol\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 686240 (2021). https://doi.org/10.3389/fimmu.2021.686240\u003c/li\u003e\n\u003cli\u003ePrasad, R.\u003cem\u003e et al.\u003c/em\u003e Plasma Microbiome in COVID-19 Subjects: An Indicator of Gut Barrier Defects and Dysbiosis. \u003cem\u003eInt J Mol Sci\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e (2022). https://doi.org/10.3390/ijms23169141\u003c/li\u003e\n\u003cli\u003eGiron, L. B.\u003cem\u003e et al.\u003c/em\u003e Markers of fungal translocation are elevated during post-acute sequelae of SARS-CoV-2 and induce NF-kappaB signaling. \u003cem\u003eJCI Insight\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e (2022). https://doi.org/10.1172/jci.insight.160989\u003c/li\u003e\n\u003cli\u003eMorais, L. H., Schreiber, H. L. t. \u0026amp; Mazmanian, S. K. The gut microbiota-brain axis in behaviour and brain disorders. \u003cem\u003eNat Rev Microbiol\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 241-255 (2021). https://doi.org/10.1038/s41579-020-00460-0\u003c/li\u003e\n\u003cli\u003eAncuta, P.\u003cem\u003e et al.\u003c/em\u003e Microbial translocation is associated with increased monocyte activation and dementia in AIDS patients. \u003cem\u003ePLoS One\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, e2516 (2008). https://doi.org/10.1371/journal.pone.0002516\u003c/li\u003e\n\u003cli\u003eChen, X., Wei, J., Zhang, Y., Zhang, Y. \u0026amp; Zhang, T. Crosstalk between gut microbiome and neuroinflammation in pathogenesis of HIV-associated neurocognitive disorder. \u003cem\u003eJ Neurol Sci\u003c/em\u003e \u003cstrong\u003e457\u003c/strong\u003e, 122889 (2024). https://doi.org/10.1016/j.jns.2024.122889\u003c/li\u003e\n\u003cli\u003eLeclerc, L., Poudrier, J., Power, C., Lam, G. Y. \u0026amp; Falcone, E. L. Intestinal barrier compromise, viral persistence, and immune dysregulation converge on neurological sequelae in Long COVID. \u003cem\u003eFront Aging Neurosci\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, 1744415 (2025). https://doi.org/10.3389/fnagi.2025.1744415\u003c/li\u003e\n\u003cli\u003eXie, J., Haesebrouck, F., Van Hoecke, L. \u0026amp; Vandenbroucke, R. E. Bacterial extracellular vesicles: an emerging avenue to tackle diseases. \u003cem\u003eTrends Microbiol\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1206-1224 (2023). https://doi.org/10.1016/j.tim.2023.05.010\u003c/li\u003e\n\u003cli\u003eVillard, A., Boursier, J. \u0026amp; Andriantsitohaina, R. Microbiota-derived extracellular vesicles and metabolic syndrome. \u003cem\u003eActa Physiol (Oxf)\u003c/em\u003e \u003cstrong\u003e231\u003c/strong\u003e, e13600 (2021). https://doi.org/10.1111/apha.13600\u003c/li\u003e\n\u003cli\u003eVillard, A., Boursier, J. \u0026amp; Andriantsitohaina, R. Bacterial and eukaryotic extracellular vesicles and nonalcoholic fatty liver disease: new players in the gut-liver axis? \u003cem\u003eAm J Physiol Gastrointest Liver Physiol\u003c/em\u003e \u003cstrong\u003e320\u003c/strong\u003e, G485-G495 (2021). https://doi.org/10.1152/ajpgi.00362.2020\u003c/li\u003e\n\u003cli\u003eFizanne, L.\u003cem\u003e et al.\u003c/em\u003e Faeces-derived extracellular vesicles participate in the onset of barrier dysfunction leading to liver diseases. \u003cem\u003eJ Extracell Vesicles\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e12303 (2023). https://doi.org/10.1002/jev2.12303\u003c/li\u003e\n\u003cli\u003eFasano, A. Zonulin and its regulation of intestinal barrier function: the biological door to inflammation, autoimmunity, and cancer. \u003cem\u003ePhysiol Rev\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, 151-175 (2011). https://doi.org/10.1152/physrev.00003.2008\u003c/li\u003e\n\u003cli\u003eKraeuter, A. K., Guest, P. C. \u0026amp; Sarnyai, Z. The Y-Maze for Assessment of Spatial Working and Reference Memory in Mice. \u003cem\u003eMethods in molecular biology\u003c/em\u003e \u003cstrong\u003e1916\u003c/strong\u003e, 105-111 (2018). \u003c/li\u003e\n\u003cli\u003eMendes de Almeida, V.\u003cem\u003e et al.\u003c/em\u003e Gut microbiota from patients with COVID-19 cause alterations in mice that resemble post-COVID symptoms. \u003cem\u003eGut Microbes\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 2249146 (2023). https://doi.org/10.1080/19490976.2023.2249146\u003c/li\u003e\n\u003cli\u003eForte, D.\u003cem\u003e et al.\u003c/em\u003e Circulating extracellular particles from severe COVID-19 patients show altered profiling and innate lymphoid cell-modulating ability. \u003cem\u003eFront Immunol\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 1085610 (2023). https://doi.org/10.3389/fimmu.2023.1085610\u003c/li\u003e\n\u003cli\u003eNair, S.\u003cem\u003e et al.\u003c/em\u003e Genomic communication via circulating extracellular vesicles and long-term health consequences of COVID-19. \u003cem\u003eJ Transl Med\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 709 (2023). https://doi.org/10.1186/s12967-023-04552-2\u003c/li\u003e\n\u003cli\u003eJones, E. J.\u003cem\u003e et al.\u003c/em\u003e The Uptake, Trafficking, and Biodistribution of Bacteroides thetaiotaomicron Generated Outer Membrane Vesicles. \u003cem\u003eFront Microbiol\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 57 (2020). https://doi.org/10.3389/fmicb.2020.00057\u003c/li\u003e\n\u003cli\u003eDoyon-Laliberte, K.\u003cem\u003e et al.\u003c/em\u003e Excess BAFF Alters NR4As Expression Levels and Breg Function of Human Precursor-like Marginal Zone B-Cells in the Context of HIV-1 Infection. \u003cem\u003eInt J Mol Sci\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e (2022). https://doi.org/10.3390/ijms232315142\u003c/li\u003e\n\u003cli\u003eAranguren, M.\u003cem\u003e et al.\u003c/em\u003e Subclinical Atherosclerosis Is Associated with Discrepancies in BAFF and APRIL Levels and Altered Breg Potential of Precursor-like Marginal Zone B-Cells in Long-Term HIV Treated Individuals. \u003cem\u003eVaccines (Basel)\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e (2022). https://doi.org/10.3390/vaccines11010081\u003c/li\u003e\n\u003cli\u003eBecker-Merok, A., Nikolaisen, C. \u0026amp; Nossent, H. C. B-lymphocyte activating factor in systemic lupus erythematosus and rheumatoid arthritis in relation to autoantibody levels, disease measures and time. \u003cem\u003eLupus\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 570-576 (2006). https://doi.org/10.1177/0961203306071871\u003c/li\u003e\n\u003cli\u003eChu, V. T.\u003cem\u003e et al.\u003c/em\u003e Systemic activation of the immune system induces aberrant BAFF and APRIL expression in B cells in patients with systemic lupus erythematosus. \u003cem\u003eArthritis \u0026amp; Rheumatism\u003c/em\u003e \u003cstrong\u003e60\u003c/strong\u003e, 2083-2093 (2009). https://doi.org/https://doi.org/10.1002/art.24628\u003c/li\u003e\n\u003cli\u003eSteri, M.\u003cem\u003e et al.\u003c/em\u003e Overexpression of the Cytokine BAFF and Autoimmunity Risk. \u003cem\u003eN Engl J Med\u003c/em\u003e \u003cstrong\u003e376\u003c/strong\u003e, 1615-1626 (2017). https://doi.org/10.1056/NEJMoa1610528\u003c/li\u003e\n\u003cli\u003eLesley, R.\u003cem\u003e et al.\u003c/em\u003e Reduced competitiveness of autoantigen-engaged B cells due to increased dependence on BAFF. \u003cem\u003eImmunity\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 441-453 (2004). https://doi.org/10.1016/s1074-7613(04)00079-2\u003c/li\u003e\n\u003cli\u003eDoyon-Lalibert\u0026eacute;, K.\u003cem\u003e et al.\u003c/em\u003e Excess BAFF May Impact HIV-1-Specific Antibodies and May Promote Polyclonal Responses Including Those from First-Line Marginal Zone B-Cell Populations. \u003cem\u003eCurr Issues Mol Biol\u003c/em\u003e \u003cstrong\u003e46\u003c/strong\u003e, 25-43 (2023). https://doi.org/10.3390/cimb46010003\u003c/li\u003e\n\u003cli\u003eSteimle, A., Autenrieth, I. B. \u0026amp; Frick, J.-S. Structure and function: Lipid A modifications in commensals and pathogens. \u003cem\u003eInternational Journal of Medical Microbiology\u003c/em\u003e \u003cstrong\u003e306\u003c/strong\u003e, 290-301 (2016). https://doi.org/https://doi.org/10.1016/j.ijmm.2016.03.001\u003c/li\u003e\n\u003cli\u003eBehrouzi, A.\u003cem\u003e et al.\u003c/em\u003e Comparative study of pathogenic and non-pathogenic Escherichia coli outer membrane vesicles and prediction of host-interactions with TLR signaling pathways. \u003cem\u003eBMC Res Notes\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 539 (2018). https://doi.org/10.1186/s13104-018-3648-3\u003c/li\u003e\n\u003cli\u003eFastQC: a quality control tool for high throughput sequence data (2010).\u003c/li\u003e\n\u003cli\u003eFranzosa, E. A.\u003cem\u003e et al.\u003c/em\u003e Species-level functional profiling of metagenomes and metatranscriptomes. \u003cem\u003eNat Methods\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 962-968 (2018). https://doi.org/10.1038/s41592-018-0176-y\u003c/li\u003e\n\u003cli\u003eManghi, P.\u003cem\u003e et al.\u003c/em\u003e MetaPhlAn 4 profiling of unknown species-level genome bins improves the characterization of diet-associated microbiome changes in mice. \u003cem\u003eCell Reports\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 112464 (2023). https://doi.org/https://doi.org/10.1016/j.celrep.2023.112464\u003c/li\u003e\n\u003cli\u003eMcMurdie, P. J. \u0026amp; Holmes, S. phyloseq: an R package for reproducible interactive analysis and graphics of microbiome census data. \u003cem\u003ePLoS One\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e61217 (2013). https://doi.org/10.1371/journal.pone.0061217\u003c/li\u003e\n\u003cli\u003eWhitmore, N. \u003cem\u003eR for Conservation and Development Projects \u003cem\u003eA Primer for Practitioners\u003c/em\u003e. (Chapman and Hall / CRC Press, 2022).\u003c/em\u003e\u003c/li\u003e\n\u003cli\u003eRobinson, M. D., McCarthy, D. J. \u0026amp; Smyth, G. K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 139-140 (2010). https://doi.org/10.1093/bioinformatics/btp616\u003c/li\u003e\n\u003cli\u003eCao, Y.\u003cem\u003e et al.\u003c/em\u003e microbiomeMarker: an R/Bioconductor package for microbiome marker identification and visualization. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 4027-4029 (2022). https://doi.org/10.1093/bioinformatics/btac438\u003c/li\u003e\n\u003cli\u003eBolger, A. M., Lohse, M. \u0026amp; Usadel, B. Trimmomatic: a flexible trimmer for Illumina sequence data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 2114-2120 (2014). https://doi.org/10.1093/bioinformatics/btu170\u003c/li\u003e\n\u003cli\u003eMasella, A. P., Bartram, A. K., Truszkowski, J. M., Brown, D. G. \u0026amp; Neufeld, J. D. PANDAseq: paired-end assembler for illumina sequences. \u003cem\u003eBMC Bioinformatics\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 31 (2012). https://doi.org/10.1186/1471-2105-13-31\u003c/li\u003e\n\u003cli\u003eMlaga, K. D.\u003cem\u003e et al.\u003c/em\u003e HCK and ABAA: A Newly Designed Pipeline to Improve Fungi Metabarcoding Analysis. \u003cem\u003eFront Microbiol\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 640693 (2021). https://doi.org/10.3389/fmicb.2021.640693\u003c/li\u003e\n\u003cli\u003eFu, L., Niu, B., Zhu, Z., Wu, S. \u0026amp; Li, W. CD-HIT: accelerated for clustering the next-generation sequencing data. \u003cem\u003eBioinformatics\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 3150-3152 (2012). https://doi.org/10.1093/bioinformatics/bts565\u003c/li\u003e\n\u003cli\u003eLu, J. \u0026amp; Salzberg, S. L. Ultrafast and accurate 16S rRNA microbial community analysis using Kraken 2. \u003cem\u003eMicrobiome\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 124 (2020). https://doi.org/10.1186/s40168-020-00900-2\u003c/li\u003e\n\u003cli\u003eDixon, P. VEGAN, a package of R functions for community ecology. \u003cem\u003eJournal of vegetation science\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 927-930 (2003). https://doi.org/10.1111/j.1654-1103.2003.tb02228.x\u003c/li\u003e\n\u003cli\u003eSeibenhener, M. L. \u0026amp; Wooten, M. C. Use of the Open Field Maze to measure locomotor and anxiety-like behavior in mice. \u003cem\u003eJ Vis Exp\u003c/em\u003e, e52434 (2015). https://doi.org/10.3791/52434\u003c/li\u003e\n\u003cli\u003eChen, C. X.\u003cem\u003e et al.\u003c/em\u003e A Multistep Workflow to Evaluate Newly Generated iPSCs and Their Ability to Generate Different Cell Types. \u003cem\u003eMethods Protoc\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e (2021). https://doi.org/10.3390/mps4030050\u003c/li\u003e\n\u003cli\u003eDorion, M.-F.\u003cem\u003e et al.\u003c/em\u003e An adapted protocol to derive microglia from stem cells and its application in the study of CSF1R-related disorders. \u003cem\u003eMolecular Neurodegeneration\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 31 (2024). https://doi.org/10.1186/s13024-024-00723-x\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8876425/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8876425/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Post COVID-19 condition (Long COVID, LC) is a heterogenous post-infectious condition frequently accompanied by persistent neurological symptoms, but its mechanisms remain unclear1-4. Here we identify gut microbiota-derived extracellular vesicles (GMEVs) as effectors linking LC-associated intestinal dysbiosis to systemic and neuroinflammation. In a longitudinal, deeply phenotyped cohort, individuals with LC and neurological symptoms (LC-Neuro) show a distinct intestinal microbiome profile that persists over time. Transplantation of LC-Neuro microbiota into germ-free mice disrupts intestinal barrier integrity and induces neurobehavioral alterations and neuroinflammation. GMEVs isolated from individuals with LC activate intestinal epithelial cells, macrophages and induced pluripotent stem cell-derived microglia in vitro, engaging inflammasome signalling, impairing epithelial barrier function and promoting inflammatory cytokine production. These effects are strongest for LC-Neuro-derived GMEVs in gut epithelium and macrophage models, whereas microglial activation is observed across LC-derived GMEVs. Oral administration of LC-Neuro GMEVs to conventional mice is sufficient to induce intestinal inflammation and systemic immune activation, accompanied by neurobehavioral changes and neuroinflammation. Together, these findings implicate GMEVs as mediators of gut-brain axis dysfunction and provide a mechanistic framework linking intestinal dysbiosis to neurological sequelae in LC.","manuscriptTitle":"Microbiota-derived extracellular vesicles mediate gut-brain axis dysfunction in long COVID","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-27 07:36:34","doi":"10.21203/rs.3.rs-8876425/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-immunology","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"ni","sideBox":"Learn more about [Nature Immunology](http://www.nature.com/ni/)","snPcode":"","submissionUrl":"","title":"Nature Immunology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"0eff4525-7313-43d9-8748-cb170dea43e7","owner":[],"postedDate":"March 27th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":63398223,"name":"Biological sciences/Microbiology/Microbial communities/Microbiome"},{"id":63398224,"name":"Biological sciences/Immunology/Inflammation/Chronic inflammation"},{"id":63398225,"name":"Health sciences/Diseases/Infectious diseases/Viral infection"},{"id":63398226,"name":"Biological sciences/Immunology/Mucosal immunology"}],"tags":[],"updatedAt":"2026-03-27T07:36:35+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-27 07:36:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8876425","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8876425","identity":"rs-8876425","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}