Cerebral ischaemic stroke results in altered mucosal antibody responses and host-commensal microbiota interactions

Abstract

Stroke is a devastating neurological event with a high risk of mortality that results in long-term sequalae that extend beyond the central nervous system. Notably these include gastrointestinal dysfunction and altered composition of the commensal microbiota in both patients and mouse models, which have been suggested to contribute to secondary infection and poor clinical outcomes following stroke. Strikingly, changes in commensal microbial community composition occur rapidly following stroke and correlate with disease severity. Despite these observations, the underpinning mechanisms that drive perturbation of the microbiota post-stroke remain poorly understood. The gastrointestinal tract is home to a complex network of tissue-resident immune cells that maintain homeostatic interactions with commensal microbes and prevent bacterial-driven inflammation. Here we demonstrate mice subjected to ischaemic stroke exhibit alterations in the intestinal immune system, most notably in class switched germinal centre B cells and the production of Immunoglobulin A (IgA) – a major effector response against commensal microbes. Mice lacking secretory antibodies, including IgA, exhibited a partial reversion of stroke-induced changes in microbiota composition. Together these findings demonstrate stroke is associated with dysregulation of antibody producing immune responses, which may in part explain changes in the intestinal microbiota. A mechanistic understanding of the immunological basis of stroke-associated pathologies in the periphery may open new avenues to manage the secondary complications and long-term prognosis of patients suffering from neurological disease.

1. Introduction

Stroke is a devastating neurological event that is the second leading cause of death worldwide, and a significant cause of morbidity and disability (Feigin et al., 2022). Whether ischaemic or haemorrhagic in origin, stroke leads to a plethora of primary and secondary pathologies, including motor and sensory deficits, cognitive impairment, behavioural change and mood disorders (Stroke Association, 2018; Feigin et al., 2022; Feigin et al., 2014; Iadecola and Anrather, 2011; Iadecola et al., 2020). Additionally, up to one third of patients will experience post-stroke infection, most commonly pneumonia or urinary tract infection, adversely affecting survival and recovery (Westendorp et al., 2011). As many as 50 % of stroke patients also experience gastrointestinal complications, including dysphagia, constipation, gastrointestinal bleeding, incontinence, and incomplete bowel emptying (Benakis and Liesz, 2022; Camara-Lemarroy et al., 2014; Lin et al., 2013; Schaller et al., 2006).

The gastrointestinal tract is host to a diverse community of commensal microbes that have critical roles in mammalian health and disease (Lynch and Pedersen, 2016). Increasing evidence suggests stroke is associated with rapid changes in the composition of the gastrointestinal microbiota in both mice and humans (Houlden et al., 2016; Singh et al., 2016; Stanley et al., 2018), which are associated with increased mortality and worsened neurological function (Sun et al., 2021; Xia et al., 2019; Xu et al., 2019). In experimental models of stroke in rodents, changes to the commensal microbiota are also associated with alterations in intestinal motility and loss of barrier integrity (Benakis and Liesz, 2022; Benakis et al., 2020a; Brichacek et al., 2020; Crapser et al., 2016; Delgado Jimenez and Benakis, 2021; Diaz-Marugan et al., 2023; Durgan et al., 2019; Houlden et al., 2016; Singh et al., 2016; Stanley et al., 2016; Stanley et al., 2018).

Despite these observations, the underlying factors that drive the modulation of the microbiota following stroke remain poorly understood. Within the gastrointestinal tract a layered network of tissue-resident immune cells constitutively reinforce the intestinal barrier and regulate host interactions with the intestinal microbiota (Belkaid and Hand, 2014). Increasing evidence suggests stroke is associated with systemic immune dysregulation mediated in part by aberrant sympathetic nervous signalling (Crapser et al., 2016; Houlden et al., 2016; Schulte-Herbruggen et al., 2009; Stanley et al., 2016; Tuz et al., 2022). However, the extent to which stroke causes altered mucosal immune function – or immune control of the host microbiota – remains unclear.

One major mechanism through which the intestinal immune system regulates commensal microbial composition is via the production of mucosal antibody, predominantly Immunoglobulin A (IgA) (Bunker and Bendelac, 2018; Huus et al., 2021; Macpherson et al., 2008; Pabst, 2012; Pabst and Slack, 2020; Rollenske et al., 2021). IgA regulation of the microbiota promotes mammalian health, while alterations in intestinal IgA responses can drive intestinal inflammation (Hansen et al., 2019), increased susceptibility to infection (Pabst and Slack, 2020), and metabolic dysfunction (Luck et al., 2019). Critically, IgA deficiency has been shown to result in altered gut microbiota composition in both mouse models and humans (Catanzaro et al., 2019; Rigoni et al., 2016; Suzuki et al., 2004). Here we demonstrate acute changes in the intestinal IgA response are associated with stroke in mice and provide evidence that changes in intestinal immune responses may in part explain the mechanistic basis of stroke-induced dysbiosis.

2. Results

2.1. Stroke induces changes in microbiota composition and the intestinal transcriptome

To begin to investigate whether stroke is associated with changes in intestinal immune responses and perturbation of the commensal microbiota we subjected mice to Middle Cerebral Artery Occlusion (MCAO) (Fig. 1A, Fig. S1A). In line with prior findings (Crapser et al., 2016; Stanley et al., 2016), we confirmed rapid loss of intestinal barrier integrity (Fig. 1B) – although no translocation of culturable bacteria to the liver or spleen were detected in our experiments (data not shown). We further observed marked changes in commensal microbial composition as early as 24-hours following stroke (MCAO) when compared to naïve controls (Fig. 1C, Fig. S1B), including a trend towards reduced microbial diversity (Shannon Index)(Fig. 1D), a significant decrease in Firmicutes to Bacteroides ratio (Fig. 1E), and alterations in specific microbial signatures (Fig. S1A–B), which were not observed to the same extent in mice undergoing sham surgery (Fig. 1C–E, Fig. S1B–C).

Fig. 1. Stroke alters intestinal microbiota composition and the colonic transcriptome.

A) Schematic showing experiment design: male C57BL/6 mice underwent MCAO, sham surgery, or no surgery (naïve). Faecal pellets were collected at 24 h and mice euthanised at 48 h, where colon tissue was harvested for bulk RNAseq. B) Faecal albumin concentration at 24 h, data pooled from 3 independent experiments, n = 13–14. (C-E) 16S rRNA sequencing of fecal microbiota from naïve, sham and MCAO mice. C) Bray-Curtis dissimilarity index, D) Shannon index & E) Ratio of Firmicutes:Bacteroides (F:B), data pooled from 2 independent experiments, n = 8–9. (F-J) Bulk RNA sequencing of colonic tissue from naïve, sham and MCAO mice. F) Heat map of genes with 50 highest and 50 lowest differentially expressed genes (DEGs; p < 0.01), expressed as z-scores. G) 15 curated pathways from GO-term pathway analysis. H) Volcano plot of top 1000 significant DEGs (p < 0.01) from comparison of sham and MCAO groups. Positive fold change indicates increase following stroke, negative fold change indicates decrease following stroke. (I-J) Selected exemplar genes identified from GO term pathways including I) immune system process and J) mucosal antibody regulation. Data pooled from 2 independent experiments, n = 7–9. Data presented as mean +/- SEM. Statistical tests: B, D, I + J – one-way ANOVA w/Tukey post-hoc. C= PERMANOVA, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Next, to investigate changes in intestinal physiology following stroke we first performed bulk RNA-seq on colonic gut tissue (Fig. 1F). This revealed ~1200 differentially expressed genes (DEGs; p < 0.01) between sham and stroke groups (Fig. 1F–H), with comparable changes in gene expression between the intestines of naïve and stroke mice (Fig. S1D). Accordingly, very few genes were significantly different between naïve and sham mice (Fig. S1D). GO term enrichment highlighted pathways that were significantly altered by stroke when compared to sham controls, including “positive regulation of the immune system”, “leukocyte activation” and “defence response to bacterium” (Fig. 1G), suggesting changes in the intestinal immune response may be associated with stroke. We further explored individual DEGs involved in these pathways (Fig. 1I), and additionally identified a number of genes involved in humoral immunity following stroke (Fig. 1J). Indeed, we noted that one of the most differentially expressed genes was Igkv5-43, a gene encoding an Immunoglobulin light chain, which decreased following stroke (Fig. 1H). We also subsequently identified alterations in genes associated with B cell regulation and mucosal antibody production following stroke – including Tnfsf13b (encoding BAFF) which supports B cell/plasma cell survival (Fig. 1J), and trends indicating potential changes in Igha and Jchain – genes associated with mucosal plasma cells (Fig. 1J). We further observed significant disruption of circadian clock gene expression following stroke – including Arntl and Nr1d1 (Fig. S1E) – transcriptional regulators that we have previously been shown to also be associated with regulation of the intestinal IgA response (Penny et al., 2022). Together these observations provoke the question as to whether stroke dysregulates homeostatic mucosal antibody responses to the microbiota.

2.2. Intestinal humoral immune responses are altered following stroke

Our transcriptomic analyses of whole intestinal tissue were suggestive of altered B cell and IgA responses following stroke. Previous findings suggest stroke leads to acute suppression of B cell populations in peripheral organs including the spleen (McCulloch et al., 2017), although whether this effect extends to mucosal barrier tissues and associated lymphoid tissues remains unclear. Thus, we focused our analyses on the major sites of mucosal antibody generation and secretion – namely the Peyer’s Patches (PP) and small intestinal lamina propria (Huus et al., 2021; Lycke and Bemark, 2012; Pabst and Slack, 2020). Within 48 h of surgery, we observed a significant reduction in the relative frequency and total number of germinal centre (GC) B cells (Fig. 2A–B) and IgA⁺ class-switched GC B cells (Fig. 2C–D) in the Peyer’s patches of MCAO mice, with these mice exhibiting visibly smaller germinal centres in the Peyer’s patches when compared to naïve mice or mice undergoing sham surgery alone (Fig. 2E). In contrast, naïve IgD⁺ B cells in the Peyer’s Patch were unchanged (Fig. S2A–B). Together these findings suggest stroke may disrupt homeostatic humoral responses in mucosal associated lymphoid structures.

Fig. 2. Stroke leads to dysregulation of mucosal humoral immunity.

(A-D) Flow cytometry analysis of Peyer’s patches B cell compartment. See also Supplemental Fig. 2. A) Representative flow plots of GL7⁺Fas⁺ Germinal Centre (GC) B cells, including quantification of percentage of parental gate (pre-gated as; Live CD45⁺/B220⁺CD19⁺; see also Fig. S2A), B) quantification of GC B cells as a percentage of total CD45⁺ cells, and cell number. C) Representative flow plots of IgA⁺ GC B cells, including quantification of percentage of parental gate and D) quantification as a percentage of total CD45⁺ cells, and cell number. E) Confocal immunofluorescence microscopy of IgD, GL7, CD3 and Hoechst in PP of representative naïve, sham and stroke mice. Low magnification images (left panels) who IgD and GL7 only to visualise Peyer’s patch follicular architecture. High magnification inserts (right panels) represent indicated location; white dotted line indicates the outline of GL7⁺ region of an individual B cell follicle. Representative of 3 independent experiments (n = 3 per group) and two independent experiments. (F-G) Flow cytometry analysis of small intestinal IgA⁺ lamina propria plasma cells. F) Representative flow plots and G) percentage and number. Pre-gated as; Live CD45⁺/CD3-CD5-MHCII⁺/B220^- IgD^- (see also Fig. S2C). All data were acquired at 48hrs following surgery. n = 11–14, data pooled from 3 independent experiments. Data presented as mean +/- SEM. Statistical tests; B, D & G – one-way ANOVA w/Tukey post-hoc, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

While IgA class switching takes place within the PP, the major source of IgA secreted into the intestinal lumen are lamina propria-resident IgA⁺ plasma cells (Huus et al., 2021; Pabst and Slack, 2020). In contrast to GC and IgA⁺ B cells in the Peyer’s patches, the relative proportion of CD138⁺ IgA+ plasma cells amongst total CD45⁺ immune cells in the small intestinal and colon lamina propria were unchanged following MCAO or sham surgery (Fig. 2F–G, Fig. S2C-D). Nonetheless, stroke mice exhibited a reduction in total cellularity of the small intes-tine when compared to naïve and sham surgery controls, which resulted in a small but significant reduction in IgA+ plasma cell numbers following MCAO (Fig. 2G). In contrast, we did not observe major changes in small intestinal lamina propria T cell populations following stroke, including Th17, Treg or γδ T cells (Fig. S2E–H). Together these findings suggest a suppression of intestinal humoral immunity following stroke, with a differential magnitude of effect on B cell responses versus long-lived antibody secreting plasma cells in the intestinal tract.

2.3. B cells and IgA⁺ plasma cells are increased in the brain and meninges after stroke

Recent work has implicated B cells and IgA⁺ plasma cells present within the CNS and meninges as having important roles both at steady-state and during inflammation (Brioschi et al., 2021; Clatworthy, 2021; Fitzpatrick et al., 2020; Fitzpatrick et al., 2024; Rojas et al., 2019). Thus, we asked whether changes in the humoral immune response were also observed in the brain and meninges following stroke (Fig. S3A). While B cells and IgA⁺ plasma cells were very rare within the brain parenchyma in both sham and stroke, we observed an increase in the frequency and numbers of these cells specifically in the stroke-impacted ipsilateral hemisphere when compared to the contralateral hemisphere (no infarct) (Fig. S3B-E). In line with recent evidence (Brioschi et al., 2021; Fitzpatrick et al., 2024), we also found naïve IgD⁺ B cells within the meninges of naïve animals which were increased in frequency and number in MCAO but not sham animals (Fig. 3A–C). Similarly, while IgA⁺ plasma cells were relatively rare in the meninges of naïve and sham mice, we observed an increase following stroke (Fig. 3D–F). B220⁺ IgA^- B cells and B220^-IgA⁺ plasma cells were found to be enriched (along with CD3⁺ T cells) in the superior sagittal sinus (Fig. 3G, Fig. S3F–G) and rostral-rhinal hub (Fig. 3H, Fig. S3F–G) of the meninges of stroke mice. Recent studies demonstrated migration of IgA⁺ plasma cells from the intestinal tract to the brain and meninges (Fitzpatrick et al., 2020; Rojas et al., 2019). To determine whether this may also occur during stroke we utilized a photoconvertible mouse model (KikGr)(Nowotschin and Hadjantonakis, 2009) to track IgA⁺ plasma cells following photo-conversion (KikGr-Red⁺) of a portion of the small intestine (Fig. S3H–I). In line with our previous findings, we observed an increase in the frequency of IgA⁺ plasma cells amongst immune cells in the brain in mice undergoing MCAO when compared to sham surgery (Fig. S3J). Although variable, the majority of IgA⁺ plasma cells in the brains of both sham surgery and MCAO groups exhibited photoconversion (KikGr-Red⁺) – suggestive of an intestinal origin (Fig. S3K–L). Together these findings suggest stroke not only alters humoral immunity within the gut but also within the brain’s borders, and suggest intestinal cells may contribute to the plasma cell pool within the central nervous system.

Fig. 3. B cell and IgA⁺ plasma Cell numbers increase in the meninges of mice following MCAO.

A) Representative flow plots, B) frequencies and C) numbers of naïve IgD⁺ B cells (pre-gated as CD45⁺/CD3^-CD5^-) in the meninges. D) Representative flow plots, E) frequencies and F) numbers of CD138⁺ IgA⁺ plasma cells (pre-gated as CD45⁺/CD3^-CD5^- MHCII⁺/B220^- IgD^-) in the meninges. (G-H) Immunofluorescence microscopy images of B220, IgA, CD3 and Hoechst in G) the superior sagittal sinus (left panel) and high-resolution confocal microscopy following stroke (right panel), and H) the rostral-rhinal hub of dural meninges wholemounts from representative naive and MCAO mice. See also Fig. S3. n = 7–8, data pooled from 2 independent experiments. Data presented as mean +/- SEM. Statistical tests; B, C, E & F – one-way ANOVA w/Tukey post-hoc, * p < 0.05, ** p < 0.01.

2.4. Intestinal IgA secretion is altered in stroke

While IgA⁺ plasma cells numbers in the intestinal tract were largely stable following acute stroke (Fig. 2F–G), the secretion and transport of IgA dimers into the intestinal lumen acts as a rate limiting step to determine anti-microbial responses. Thus, we next aimed to clarify the effect of stroke on secretory IgA (sIgA) present within the intestinal tract itself. Surprisingly, we found a striking increase in faecal sIgA levels following stroke in MCAO mice, but not in sham mice, as early as 24 h following stroke (Fig. 4A; normalised to total faecal protein Fig. S4A). The secretion of plasma cells can be dynamically regulated by a wide range of intrinsic and extrinsic cues (Huus et al., 2021). To determine whether the secretory capacity of IgA⁺ plasma cells was altered following stroke we sort-purified IgA⁺ CD138⁺ cells from the small intestines of naïve and MCAO mice and cultured equal numbers of cells ex vivo. Strikingly we observed significantly higher concentrations of IgA in the culture supernatants of IgA⁺ plasma cells derived from MCAO mice when compared to naïve controls (Fig. 4B) – suggesting stroke leads to increased fecal sIgA levels not through alterations in cell numbers but via alterations in plasma cell-intrinsic antibody secretion.

Fig. 4. Intestinal IgA responses against the microbiota are perturbed following stroke.

A) Faecal sIgA concentration at 24 h, data pooled across 4 independent experiments, n = 14–19. B) IgA concentrations in supernatants of sort-purified IgA⁺ plasma cells cultured ex vivo. Data pooled from 2 independent experiments, n = 6–8 per group. C) Gating strategy for bacterial flow cytometry. D) Percentage of IgA⁺ bacteria at 24 h, data pooled across 4 independent experiments, n = 13–18. (E-F) IgA-binding index (Kau Index) of selected E) Bacteroides and F) Firmicutes genera. Data pooled from 2 independent experiments, n = 7–9 per group. Data presented as mean +/- SEM. Statistical tests; A – one-way ANOVA w/Tukey post-hoc, B, E & F – Kruskal-Walis test with multiple comparisons, * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001.

IgA directly binds to commensal microbes to modulate their colonisation, growth and wider biology. Thus, next we determined whether the stroke-induced increase in sIgA was associated with changes in the proportion of antibody-bound commensal bacteria (Fig. 4C; Fig. S4B). Quantification of the global frequencies of IgA⁺ commensal microbes by bacterial flow cytometry revealed comparable proportions of total IgA-bound bacteria 24hrs post-stroke (Fig. 4D). We also determined IgG binding, which can occur in the context of reduced barrier integrity, but found no evidence of IgG labelling of bacteria (Fig. S4B-C), and no significant increases in circulating commensal-specific IgG levels in the serum (Fig. S4D). Next we performed IgA-Seq to provide a granular analysis of IgA bound commensal bacteria, which revealed significant changes in IgA binding to a number of bacteria of the Bacteroides and Firmicutes genera (Fig. 4E–F, Fig. S4E–F), suggesting IgA responses to specific commensal bacteria could be altered post-stroke.

2.5. Intestinal antibody responses are partially responsible for altered commensal microbiota

As changes to the microbiota were observed to occur rapidly following stroke (Fig. 1 & Fig. S1), and we observed a spike in both faecal sIgA and plasma cell-intrinsic IgA secretory capacity following stroke (Fig. 4A + B), we next asked whether increased luminal IgA could in part be responsible for stroke-induced alterations in the commensal microbiota. To investigate this we used IgMi mice, as reported previously (Penny et al., 2022; Waisman et al., 2007), which are unable to class-switch or secrete antibody but retain B cells and associated antibody-independent effector functions (Sahputra et al., 2018). Given IgA is the predominant antibody subtype in the intestines, and we were unable to detect bacterial-bound IgG (Fig. S4) in naïve mice or following sham or stroke, this mouse model allowed us to investigate stroke effects on microbiome composition in the absence of immunoglobulin-bound microbes (Fig. 5A). IgMi mice and littermate controls (wild type; WT) were subject to sham surgery or MCAO. Importantly, the absence of secretory antibody had no impact on the initial infarct volume or stroke-associated acute behavioural deficits (Fig. S5A-C) – indicating acute stroke severity did not significantly differ between IgMi and control mice subjected to MCAO.

Fig. 5. Stroke-associated alterations in microbiota composition are in part dependent on mucosal antibody.

A) Representative flow cytometry plots showing IgA binding to bacteria isolated from faecal pellets of IgMi mice compared to littermate controls (WT). (B-F) shotgun metagenomic sequencing of fecal microbiota isolated from IgMi and WT mice 48 h after undergoing sham or MCAO surgery. B) Bray-Curtis dissimilarity index, C) Shannon index & D) Firmicutes:Bacteroides (F:B) ratio, n = 4. Data representative of 2 independent experiments. E) Stacked bar charts representing average composition of microbiota at Phylum level in WT and IgMi mice undergoing sham or MCAO. F) Select individual genera from sham and stroke WT and IgMi mice, data pooled from 2 independent experiments, n = 7–8. Data presented as mean +/- SEM. Statistical tests; B = PERMANOVA, F = Kruskal-Wallis test with multiple comparisons, * p < 0.05, ** p < 0.01.

IgMi mice and littermate controls undergoing sham surgery had largely comparable global microbial diversity (Fig. 5B), in line with our previous findings that lack of IgA does not disrupt global community composition at homeostasis (Penny et al., 2022). As expected, microbial communities segregated differentially between stroke and sham groups in WT animals (Fig. 5B). Notably however, while the microbial signature of IgMi mice also exhibited a clear change following stroke, this group exhibited differential clustering to MCAO wild type littermates – suggestive of potential differences in microbial composition and diversity following stroke in the absence of secretory antibody (Fig. 5B).

In line with our previous observations (Fig. 1D), stroke led to a trend towards a reduction in Shannon diversity index when compared to sham controls, which occurred both in the presence and absence of IgA (Fig. 5C). Again, we observed a reduction in the Firmicutes to Bacteroides ratio in MCAO WT mice (Fig. 5D), as expected and in line with our prior observations (Fig. 1E). In contrast, the reduced ratio between these two major commensal phyla was no longer evident following stroke in IgMi mice (Fig. 5D + E). As we observed altered IgA-binding to Firmicutes and Bacteroides following stroke in wild type animals (Fig. 4), this suggests perturbations in IgA responses following stroke may in part contribute to stroke-driven changes in intestinal microbial community diversity and composition.

Finally, we further investigated the extent of stroke-induced changes to commensal microbiota composition in the presence or absence of intestinal antibody (Fig. 5E–F, Fig. S5D–F). Our analysis revealed multiple specific genera that were altered in abundance by stroke in WT MCAO mice when compared to sham controls, but which were not significantly altered following stroke in IgMi (Fig. 5F, Fig. S5D–E). Despite these specific changes we also observed changes in microbial abundance that were observed in both genotypes (Fig. S5E), as well as stroke-induced changes that only reached significance in IgMi mice, and not WT mice, following stroke (Fig. S5F). Together these findings suggest changes in mucosal IgA responses may contribute in part to microbiota alterations observed in experimental models of stroke.

3. Discussion

Alterations in gastrointestinal function and commensal microbiota composition following stroke are well described both in patients (Cui et al., 2023; Xia et al., 2019; Xu et al., 2019; Yamashiro et al., 2017; Yin et al., 2015) and animal models (Benakis et al., 2016; Brichacek et al., 2020; Houlden et al., 2016; Singh et al., 2016; Stanley et al., 2016; Stanley et al., 2018). Importantly, changes in microbial communities correlate with worsened stroke outcome and secondary complications (Benakis et al., 2016; Brichacek et al., 2020; Singh et al., 2016; Stanley et al., 2016; Tuz et al., 2022; Xia et al., 2019). Indeed, antibiotic intervention post-stroke can lead to improved outcomes (Benakis et al., 2020b; Liu et al., 2022). Despite these advances the mechanisms that underpin stroke-induced intestinal pathology and alterations to commensal microbial communities remain poorly understood.

Recent findings indicate stroke induces dysregulation of multiple intestinal immune populations (Benakis et al., 2016; Brea et al., 2021; Diaz-Marugan et al., 2023; Ge et al., 2022; McCulloch et al., 2017; Oyama et al., 2018). Intestinal tissue-resident immune cell networks are critical in regulating the commensal microbiota (Belkaid and Hand, 2014) – however whether maladapted immune responses to the microbiota mechanistically drive changes in the commensal community post stroke has been unclear. One major mechanism of host-commensal interactions is the production and transluminal secretion of IgA, which physically binds to bacteria to modulate their growth, survival, and broader biology (Bunker and Bendelac, 2018; Huus et al., 2021; Pabst and Slack, 2020; Rollenske et al., 2021). Here we demonstrate acute changes in mucosal humoral immunity, fecal IgA and the secretory capacity of IgA plasma cells following MCAO-induced stroke in mice. While the relative frequency of faecal bacteria bound by IgA did not change immediately following stroke, IgA-seq revealed discrete changes in IgA binding of distinct bacteria and we observed differential changes in microbiota composition in IgMi mice, which are unable to class switch or secrete IgA. Together these findings suggest that changes in IgA secretion and binding may in part contribute to acute post-stroke alterations in commensal communities. Despite these observations, other microbial changes occurred following stroke independent of the presence of mucosal antibody suggesting additional mechanisms contribute to changes in the microbiota. Notably, IgMi mice exhibited lesions volumes and behavioural deficits equivalent to control littermates, suggesting IgA-dependent alterations of the microbiota do not determine the initial severity of stroke pathology in this model. Together our findings support an increasing body of evidence that demonstrate an effect of stroke on host immune regulation of commensal microbes, however the impact of these changes for the long-term prognosis of stroke remain unclear.

The mechanisms through which IgA responses are perturbed post-stroke remain to be elucidated. In line the findings of a previous study (Oyama et al., 2018), we detected reductions in germinal centre B cells within the intestinal-associated Peyer’s patches following MCAO, including reductions in IgA-class switched B cells. In contrast, IgA⁺ plasma cells – which reside within the intestinal lamina propria – were not markedly reduced in number following stroke, although their intrinsic capacity to secrete IgA was altered. These data are suggestive of differential impacts of stroke on terminally differentiated antibody producing cells or between tissue sites. Moreover, while we demonstrate enhanced IgA levels in the faeces are associated with an increase in the magnitude of antibody production by tissue-resident plasma cells on a per cell basis, we cannot rule out a role for increased gut permeability in further contributing to the increased luminal IgA observed – something that could feasibly negate the requirement for active transport of IgA into the lumen via PigR. Interestingly, a parallel study demonstrates that microbiota changes persist for months following stroke (Martin et al., 2025). This study also observed disrupted Peyer’s patches responses following stroke despite rapid restoration of barrier integrity, and notably report reduced IgA binding to intestinal bacteria 3 months post stroke (Martin et al., 2025). Together these findings suggest that chronic disruption of humoral responses in Peyer’s patches following stroke may ultimately lead to consequences in the levels of IgA bound bacteria, potentially due to turnover of tissue-resident cells and a failure to replenish the plasma cell pool over longer time periods, due to the disruption of Peyer’s patch germinal centres. These insights further implicate both acute and long-term immune defects in the gastrointestinal tract following stroke.

The basis of these differential effects of stroke on Peyer’s patch B cells versus tissue-resident plasma cells at acute time points (<48 h hours) remains unclear. Plasma cells are notably long-lived and quiescent cells – in contrast to highly proliferative B cells within active germinal centres – which may make them less susceptible to cell death. A prior seminal study demonstrated the loss of splenic marginal zone B cells in MCAO mice, which was driven by noradrenaline release and suppressed B cells via the β2-adrenergic receptor (McCulloch et al., 2017). One possibility is that the quality or magnitude of innervation within the plasma cell niche could differ, or those signals may elicit differential outcomes due to intrinsic differences in plasma cell biology. Future studies are needed to further define the role of neuroimmune crosstalk in driving immunological change post-stroke.

Recent advances have demonstrated the presence of antibody secreting cells within the CNS and meninges in health and disease (Brioschi et al., 2021; Fitzpatrick et al., 2020; Probstel et al., 2020; Rojas et al., 2019). Intriguingly, these prior studies suggested IgA-producing cells found within the meninges were of intestinal origin (Fitzpatrick et al., 2020; Probstel et al., 2020; Rojas et al., 2019), indicating potential for gut-CNS migration, while more recent studies suggest dural-associated lymphoid tissues may also support local B cells responses (Fitzpatrick et al., 2024). Here we also observed B cells and IgA⁺ plasma cells present in the brain parenchyma and meninges in the acute stages following stroke – in line with previous findings that observed B cells within the brains of animals at chronic time points following stroke (Doyle et al., 2015; Ortega et al., 2020). Intriguingly, this coincided with a reduction of IgA⁺ B cells in the Peyer’s patches of the intestine raising the possibility of migration between these two sites post-stroke. Indeed, our studies with photoconvertible mice indicate possible migration of humoral immune cells from the gut to the CNS during stroke. Further studies are required to disentangle the origins of IgA-producing cells within the CNS following stroke.

Our observations, and those of prior studies, raise the question as to why humoral immune responses may expand or be recruited to the CNS in this context. It has been previously proposed that meningeal IgA production could act as a firewall for circulating microbes with potential to enter the CNS. Stroke leads to a breakdown of intestinal barrier function and increases susceptibility to secondary infection in both mouse and human. Indeed, bacteraemia is commonly reported following stroke in mice, with culturable microbes present in systemic organs including the liver, lungs and spleen (Diaz-Marugan et al., 2023; McCulloch et al., 2017; Stanley et al., 2016). Thus, it is attractive to hypothesise that an increase in antibody producing cells both within the intestinal tract and the CNS may help protect the brain from trans-locating or invading bacteria that have entered the bloodstream following stroke.

Nonetheless, our study has several key limitations that should be considered. Firstly, in many experiments we observed an effect of sham surgery when compared to naïve animals. While MCAO resulted in further marked effects on humoral immunity and microbiota composition beyond that of sham surgery alone this indicates surgery in part mediates immunomodulatory effects that could impact upon the microbiota. Notably however many of the effects induced by surgery are physiologically relevant, also occur during MCAO, and are relatable to stroke pathology. For example, hormones (e.g. glucocorticoids) and neuropeptide release results in potent modulation of the peripheral immune system and are elicited by both sham surgery and stroke, in both animal models and man. Furthermore, animals undergoing sham surgery or MCAO both exhibit a transient reduction in food and water intake, while similarly stroke patients exhibit dysphagia and reduced oral food intake (Camara-Lemarroy et al., 2014). Importantly changes in feeding can also modulate the magnitude of the IgA response (Penny et al., 2022), and perturb the intestinal microbiota (Huus et al., 2021). Thus, future studies are required to untangle the mechanisms driving stroke-associated immune sequalae in both surgical animal models and humans.

An additional, and important, consideration is that this study was performed in young adult mice and primarily focused on acute immune changes – yet stroke is primarily a disease of aging in humans and has chronic pathology and consequences. Moreover, given the association between stroke and age in human patients, future studies are required to determine the consequences of stroke-associated immune dysregulation in aged mouse cohorts, across the spectrum of acute and chronic pathologies, and as observed during stroke pathology in human cohorts.

Together this study highlights the need for further investigation into the role of the intestinal immune system in the context of stroke and has the potential to open new therapeutic avenues for the management of both the primary pathology and secondary complications arising in patients.

4. Methods

4.1. Animals

Age-matched male C57BL/6 mice were sourced from Charles River and used for studies between 8 and 12 weeks of age. IgHμγ1 mice (herein referred to as IgMi mice) were provided by Prof Ari Waisman (University of Mainz, Germany). Transgenic lines were bred in-house at the Biological Services Facility (BSF) at the University of Manchester. For experiments using transgenic lines, age- and sex-matched littermate controls were used. All mice were maintained under specific pathogen-free conditions and kept in sterilised filter-topped cages under a 12-hour light/dark cycle, individual studies were age and sex matched. Mice had access to food and water ad libitum, unless otherwise indicated. Mice were acclimatised for a period of at least 7 days before any procedures were performed. All animal experiments were conducted in accordance with the regulations stipulated in the Animal (Scientific Procedures) Act 1986. The procedures conducted for this research project were performed by individuals holding UK personal licences under project licences issued to Prof Stuart Allan. Protocols for experimental work were reviewed and approved by the University of Manchester Animal Welfare and Ethical Review Board. The ARRIVE guidelines were consulted during experimental design (Percie du Sert et al., 2020).

4.2. Stroke surgery

A modified version of the transient MCAO protocol was used to induce experimental cerebral ischaemia (Longa et al., 1989). A midline incision to the throat was made and soft tissues pulled apart to expose the left common carotid artery which was dissected free from surrounding nerves. Two ligatures were made around the left common carotid artery (spaced approximately 3 mm apart) before it bifurcates into the left internal and external carotid arteries, which were then also isolated and tied off. A small incision was made in the common carotid artery in the region between the two ligatures, then a nylon mono-filament (diameter 180 μm) coated with silicon was inserted into the common carotid artery, advanced into the middle cerebral artery via the internal carotid artery by 10 mm and left in place for 30 min. The filament was then withdrawn and the wound closed with 6–0 Vicryl sutures (Ethicon). For sham surgeries ligatures were made but the carotid incision and filament insertion was omitted. Post-operative care was given as follows: Buprenorphine at 0.05 mg/kg and 500 μl saline was injected subcutaneously for analgesia and fluid replacement respectively, and the animals were placed in a heated cabinet during recovery. Mice were provided with floor-fed wet mash and given 0.05 mg/kg Buprenorphine and 300 μl saline daily until sacrifice. Mice were randomised to intervention groups using block randomisation, to avoid cage effects. For microbiota-focused experiments, mice were housed with others of the same group post-surgery (i.e. stroke mice were only housed with other stroke mice) to avoid horizontal transfer of microbiota. Blinding to experimental groups and/or genotypes was only done where specified (i.e. neurological deficit scoring) due to need to track and separate genotypes during surgery. Approximately 7 % of all animals used in this study were prematurely euthanised due to i) exhibiting a decline in condition post-surgery that approached the severity limits of our license, or ii) a failure to recover from anaesthesia during surgery. Additionally, post hoc exclusion of data from n = 5 mice occurred due to a failure to confirm stroke, assessed via either front limb symmetry by tail suspension, and/or absence of brain lesion as shown by either MRI or Cresyl violet staining.

4.3. Behavioural and pathological readouts

Where indicated, post-stroke neurological deficits were tested at day 1 after surgery using a 28-point scale adapted from Clark et al. (1998). Seven criteria were assessed; in brief, these were body asymmetry, gait, climbing ability, circling behaviour, front limb asymmetry, compulsory circling and response to whisker touch. Neurological assessment was carried out by an experienced researcher who was blinded to experimental groups. In some cases, to accurately quantify ischaemic lesion volume mice were scanned by T2-weighted MRI at day 1 post-surgery. MRI scans were acquired using a 7 T horizontal bore magnet (Agilent Technologies, CA, USA) connected to a BrukerAvance III console (Bruker Biospin, MA, USA) using a surface transmit-receive coil. Mice underwent a short localisation scan, followed by a 19-minute T2 acquisition in which 20 coronal scans of 1 mm depth were acquired. Lesion volumes were measured in Fiji (ImageJ, NIH) by calculating the oedema ratio (area of contralateral hemisphere/ipsilateral) and then multiplying this by the lesion area to give an oedema-corrected lesion volume (Nouraee et al., 2019).

4.4. Faecal pellet collection

Fresh faecal pellets were obtained either by placing mice in clean cages without bedding and waiting for defecation, or removed directly from the colon at necropsy. In each case care was taken to ensure collection tools were sterilised between mice and no contamination of samples occurred. To separate the bacteria and supernatants for ELISA, faecal pellets were resuspended in sterile PBS at a concentration of 1 μl/mg and incubated on ice for 20 min. Pellets were homogenised for 30 s at 4.0 m/s on a tissue homogeniser (Fastprep 24, MP Biomedical) then centrifuged for 5 min at 200g to remove debris. Supernatants were filtered through a 70 μm cell strainer and centrifuged again at 8000g for 5 min to pellet the bacteria. Supernatant and bacterial pellets were stored at −80 °C until analysis.

4.5. ELISA

Faecal albumin, IgA and IgG were quantified in normalised faecal supernatants, serum or ex vivo culture supernatants using either ELISA Quantitation Sets (Bethyl Laboratories) or Mouse IgA ELISA kits (Invitrogen) following manufacturer’s instructions. To measure commensal-specific IgG ELISA plates were coated with 5ug/ml commensal antigen (previously described in (Hepworth et al., 2015) overnight at 4 °C, and subsequently washed and blocked with 1 % BSA in PBS for 90 min at room temperature. Following incubation with samples and further washes, IgG was detected using an alkaline phosphatase-conjugated IgG detection antibody (Southern Biotec), and developed with p-nitrophenyl phosphate.

4.6. Bicinchoninic acid (BCA) assay

To measure total protein in faecal pellet supernatants, the Pierce™ BCA Protein Assay Kit (ThermoFisher Scientific) was used according to manufacturer’s instructions. In brief, standards and sample replicates were pipetted onto a 96-well plate, after which a working solution was added at a sample:solution ratio of 1:20. Plates were then covered and incubated at 37 °C for 30 min before cooling and reading absorbance on a spectrophotometer (Versamax, Molecular Devices) at an optical density of 562 nm.

4.7. Bacterial flow cytometry & IgA-seq

Samples for bacterial flow cytometry and IgA-seq were processed from faecal pellets as described above. For flow cytometry, pellets were stained with PE-IgA and SB600-IgK at 1:200 and SYTO-60 at 1:600 in FACS buffer for 30 min at 4 °C. Samples were washed, resuspended in PBS and acquired on a BD LSRFortessaTM using DIVA software (BD Biosciences). Flow cytometry data was analysed and flow plots were produced in FlowJo V10. For IgA-seq, pellets were stained in the same manner, and after washing were resuspended in PBS. A 200 μl aliquot was taken and frozen at –80 °C (“total bacteria” sample) alongside a 50 μl sample for flow cytometry analysis. The remaining bacteria were incubated with anti-PE beads (Miltenyi) for 30 min at 4 °C before centrifugation and washing with FACS buffer. The sample was then passed through an LS column (Miltenyi) placed on a QuadroMACSTM Separator (Miltenyi) to isolate IgA⁺ labelled bacteria. The IgA-run-through was collected and the bound fraction was removed from the magnetic field and eluted (IgA⁺ sample). Aliquots were taken to validate successful isolation via flow cytometry analysis and the remaining sample was pelleted and frozen at –80 °C until processed for genomic DNA extraction. All samples underwent 16S sequencing as describe below. Data were prefiltered in excel then analysed using R studio and GraphPad. IgA-binding index (Kau index (Kau et al., 2015)) was calculated for each taxon present in an individual sample.

4.8. 16S rRNA sequencing

Whole faecal pellets were obtained as described above. Pellets were collected either into sterile tubes on ice and frozen at −20 °C until DNA extraction and 16S rRNA-seq by the Centre for Genomic Research (CGR), Liverpool, UK). Bacterial genomic DNA was isolated with the DNeasy PowerLyzer PowerSoil kit (Qiagen) following manufacturer’s instructions. Briefly, faecal pellets were homogenized and bacteria isolated as described above. The bacterial pellet was then transferred to a PowerBead Tube and homogenized again at 4 m/s for 30 sec. Protein was precipitated and inhibitors were removed before loading the sample into an MB Spin Column to bind the DNA before eluting into an Eppendorf. All steps were done in a tissue culture hood with sterile equipment and unopened kit reagents to avoid contamination of samples.

Prior to sequencing, pre-amplification of the V3/V4 region of 16S rRNA was performed by PCR in triplicate using 2xKAPA HiFi Hot Start ReadyMix (Roche) using primer pairs containing adaptor sequences for down-stream use on Illumina platforms, as follows:

16S Amplicon PCR Forward Primer: 5′ TCGTCGGCAGCGTCA-GATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG 3′.

16S Amplicon PCR Reverse Primer: 5′ GTCTCGTGGGCTCGGA-GATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC 3′. Successful amplification was confirmed by running samples on a 1.5 % agarose gel and confirmation of a single PCR product of ~ 380 bp. PCR products were shipped on dry ice to the CGR for downstream sequencing and analysis.

4.9. Shotgun metagenomic sequencing

Shotgun sequencing was carried out by Transnetyx. In brief 2–3 faecal pellets were transferred to barcoded collection tubes containing DNA stabilisation buffer provided by Transnetyx. Samples were shipped to Transnetyx where DNA extraction was performed using the Qiagen DNeasy 96 PowerSoil QIAcube HT extraction kit. Library assembly was done using the KAPA hyperplus library preparation protocol, and sequenced with the Illumina NovaSeq instrument and protocol at a depth of 2 million (2×150 bp pair reads). Unique dual indexed (UDI) adapters were used to ensure that reads and/or organisms were not mis-assigned. Raw data (FASTQ files) were analyzed using the One Codex database, before exporting and analyzing in R Studio.

4.10. Analysis of microbiome sequences

FASTQ sequences were uploaded to the One Codex database of whole microbial reference genomes and aligned using k-mers, where k = 31. Based on the relative frequency of unique k-mers in the sample, probable sequencing or reference genome artifacts were filtered out of the sample. Then, relative abundance of each species was estimated based on the depth and coverage of sequencing across every available reference genome. Relative abundances were used to create stacked bar graphs of phyla and genera. PCoA were produced in R Studio. Only genera forming >1 % of total abundance were used for generating bar graphs; genera with abundance below this are shown as “other”.

4.11. RNA extraction

Tissue sections approximately 1 cm long were dissected from the distal colon and flash frozen on dry ice. Samples were homogenized in 40 μl of RLT buffer (Qiagen) in lysing matrix tubes (MP Biomedical) for 30 sec at 4.0 m/s on a tissue homogeniser (Fastprep 24, MP Biomedical). Total RNA was extracted using either an RNeasy mini kit (Qiagen) or the Norgen Single Cell RNA Purification Kit (Norgen Biotek Corp, Canada) following the manufacturer’s instructions. Sample RNA concentration was measured either using a Nanodrop for Real-Time PCR (qPCR), or using the QubitTM RNA HS Assay Kit (Thermo Fisher Scientific) and protocol for bulk-RNA sequencing.

4.12. Bulk RNA-seq

Small intestine and colon RNA samples were frozen and shipped on dry ice to Novogene (Cambridge, UK) for sequencing. Novogene carried out in-house quality control and normalisation of raw FASTQ files. Data was analysed using the DESeq package in R Studio. Principal component analysis (PCA) plots were generated in R Studio. Z-scores were calculated and plotted as heatmaps in R studio. Fold change was calculated and used to generate volcano plots in GraphPad Prism v8. GO term analysis was carried out using the Gene Ontology Resource (Ashburner et al., 2000; Gene Ontology Consortium, 2021).

4.13. Intestinal lamina propria cell isolation

To isolate the lamina propria lymphocytes, the small intestine and colon were dissected, associated fat removed and the tissue was cut open longitudinally, washed in PBS to remove luminal contents, and placed in 15 ml PBS on ice. To fully remove luminal contents, tissue was vortexed and washed with PBS 3 times. Tissue was then placed in 15 ml of stripping buffer (1 mM EDTA, 1 mM DTT and 5 % FCS) for 10 min at 37 °C on a tube rotator (Fisher Scientific), followed by a further 20 min incubation with rotation in fresh strip buffer to removed epithelial cells and intra-epithelial lymphocytes. The remaining tissue was washed with PBS and then incubated in 10 ml of digestion buffer (0.1 mg/ml collagenase/dispase (Roche) OR 1 mg/ml collagenase D (Roche, Switzerland), and 20 μg/ml DNase I (Sigma-Aldrich, St Louis, USA) for 45 min at 37 °C on a tube rotator (Fisher Scientific. The resulting tissue and supernatant were passed through a 70 μm cell strainer (Corning, New York, USA) and centrifuged to isolate lamina propria lymphocytes. Cell counts were obtained using a Casy TT counter (Roche Innovatis, Germany).

4.14. Peyer’s patch cell isolation

Between 4–6 Peyer’s patches were excised from the small intestine at necropsy and placed directly in digestion media containing 0.16 mg/ml Liberase (Roche, Switzerland) and 40 μg/ml DNase (Sigma-Aldrich, St Louis, USA) on ice. Peyer’s patches were then incubated in the digestion media at 37 °C for 30 min before mashing through a 70 μm cell strainer (Corning, New York, USA) into complete media.

4.15. Brain and meninges cell isolation

Whole brains (including leptomeninges and choroid plexus) were placed in RPMI 1640 containing 2 mM L-glutamine, 100 units/ml penicillin, 100 μg/ml streptomycin, 10 % foetal calf serum (FCS)), collagenase D (10 μl/ml) and DNase (10 μl/ml)) and cut finely with scissors before incubating for 30 min at 37 °C with agitation. Resulting suspensions were filtered through a 70 μm cell strainer (Corning, New York, USA) and centrifuged. The supernatant was discarded, the pellet resuspended, and centrifuged on a Percoll gradient. The uppermost layer containing myelin was carefully removed with a Pasteur pipette and remaining suspension diluted and mixed thoroughly in media, then centrifuged to isolate the leukocytes.

The meninges were peeled from the inside of the skull and placed in RPMI 1640 containing 500 U/ml DNase I and 140 U/ml Collagenase 8 (Sigma Aldrich) and incubated for 20 min at 37 °C with agitation. The resulting suspension was filtered through a 70 μl cell strainer (Corning, New York, USA) and centrifuged.

4.16. Flow cytometry

Single cell suspensions were resuspended in FACS buffer (PBS with 4 % FCS and 1 mM EDTA) containing a surface stain antibody cocktail and Fixable Aqua cell viability dye (ThermoFisher) and incubated for 30 min at 4 °C in the dark. To enable detection of transcription factors and cytokines, cells were resuspended in FoxP3 fix/perm buffer (eBio-science) for 30 min then centrifuged and washed with FoxP3 1x perm buffer (eBioscience). Intracellular antibody cocktails were made up in FoxP3 1x perm buffer and applied to cells for 30 min at 4 °C in the dark. Fluorescence minus one (FMO) controls were used to help with gating for cytokines. After staining, samples were washed, resuspended in FACS buffer and acquired on a BD LSRFortessaTM using DIVA software (BD Biosciences). Flow cytometry data was analysed and flow plots were produced in FlowJo V10. All antibodies listed in Table 1 of the Supplemental Material.

Table 1. List of flow cytometry antibodies.

Target	Clone	Conjugate
CD4	GK1.5	BUV395
CD19	1D3	BV421
MHCII	M5/114.15.2	eFluor 45044
FoxP3	FJK-16s	FITC
CD11b	M1/70	FITC
CD11b	M1/70	APC-e Fluor 780
CD11c	N418	APC-e Fluor 780
Fas	15A7	FITC
IgA	mA-6E1	PE
RORgt	B20	PE
CD25	PC61	SB600
B220	RA3-6B2	SB600
CD3	14s-2c11	PE-TxRed
CD3	14s-2c11	PerCP-Cy5.5
CD5	53-7.3	PerCP-Cy5.5
NK1.1	PK136	PerCP-Cy5.5
IgD	11-26c.2a	PE-TxRed
CD45	30-F11	BV650
GL7	GL7	APC
B220	RA3-6B2	APC-e Fluor 780
CD138	281-2	PE/Cy7

4.17. Cell sorting and ex vivo IgA secretion assay

IgA⁺ plasma cells were sorted from small intestinal lamina propria as Live CD45⁺/CD3^- CD5^- NK1.1^- MHCII^lo/+/B220-IgD^-/CD138⁺ IgA⁺ using a FACSAria II (BD Biosciences). Experimental replicates were derived from individual mice, cell numbers were normalised, and 10,000 cells were cultured overnight in RPMI 1640 media containing 10 % FCS, 10 ng/ml IL-6 (PeproTech, USA) and 200 ng/ml BAFF (Bio-Legend, UK) in a 96 well plate.

4.18. Whole mount meninges imaging

Mice were transcardially perfused with sterile saline under anaesthesia. Skull caps with the intact dural meninges were collected and incubated in 4 % PFA at 4 °C for 6 h. The meninges were then carefully removed from the interior surface of the skull and transferred to a 24-well plate for staining with the following: fluorescent-conjugated anti-bodies AF488-B220, PE-CD138, AF647-CD3, AF555-IgA and APC-IgA in 2 % BSA 0.2 % Triton and 2 % donkey serum in PBS. Meninges were incubated in the dark on a shaker overnight at 4 °C, followed by washing with PBS then DAPI was added for 5 min. Meninges were washed again with PBS before mounting. To mount, meninges were floated in a petri dish containing PBS and carefully mounted onto SuperFrostTM Plus slides (Thermo Fisher Scientific) and left to dry for 1–2 h. Once dry 1–2 drops of ProLongTM Gold antifade mountant (Invitrogen) were added and slides were coverslipped and allowed to dry in the dark before imaging. Whole-mount wide-field microscopy images were collected on an Olympus BX63 upright microscope using a 10x/0.4 UApo/340 objective and captured and white-balanced using a DP80 camera (Olympus) in colour mode through CellSens Dimension v1.16 (Olympus). Specific band pass filter sets for DAPI, FITC, Texas red and CY5 were used to prevent bleed through from one channel to the next. Confocal micro-scopy images were collected using a Leica SP8 inverted microscope using a 40x objective. Images were processed and analysed using Fiji (ImageJ, NIH).

4.19. Data analysis and statistics

All raw data were processed using Microsoft Excel and analysis was carried out in GraphPad Prism, unless otherwise specified. All flow cytometry data was analysed and flow plots were produced in FlowJo V10. Imaging data was processed and analysed using ImageJ (NIH). Experimental schematics were produced with Biorender. Data are reported as mean ± SEM unless otherwise stated, with details of specific statistical tests indicated in figure legends. For all data, p < 0.05 = *; p < 0.01 = **; p < 0.001 = ***; p < 0.0001 = ****.

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbi.2025.106184.

Data availability

Data will be made available on request.