Advanced age promotes colonic dysfunction and gut‐derived lung infection after stroke

Abstract Bacterial infection a leading cause of death among patients with stroke, with elderly patients often presenting with more debilitating outcomes. The findings from our retrospective study, supported by previous clinical reports, showed that increasing age is an early predictor for developing fatal infectious complications after stroke. However, exactly how and why older individuals are more susceptible to infection after stroke remains unclear. Using a mouse model of transient ischaemic stroke, we demonstrate that older mice (>12 months) present with greater spontaneous bacterial lung infections compared to their younger counterparts (7–10 weeks) after stroke. Importantly, we provide evidence that older poststroke mice exhibited elevated intestinal inflammation and disruption in gut barriers critical in maintaining colonic integrity following stroke, including reduced expression of mucin and tight junction proteins. In addition, our data support the notion that the localized pro‐inflammatory microenvironment driven by increased tumour necrosis factor‐α production in the colon of older mice facilitates the translocation and dissemination of orally inoculated bacteria to the lung following stroke onset. Therefore, findings of this study demonstrate that exacerbated dysfunction of the intestinal barrier in advanced age promotes translocation of gut‐derived bacteria and contributes to the increased risk to poststroke bacterial infection.

. Specifically, the adverse biological effects of normal aging, such as cell senescence, low-grade systemic inflammation and decline in immune function, can impend on events critical to recovery after stroke and contribute to the poor prognosis in the elderly (Licastro et al., 2005;Ritzel et al., 2018;. Despite its well-recognized primary effects on the brain, a major cause of death after stroke is infection: a poststroke complication that has received increasing attention for its large clinical implications (Shim & Wong, 2018). In fact, more than 30% of infected patients with stroke die of infection as a secondary complication within a week of stroke onset, with infection of the respiratory (bacterial pneumonia) and urinary tracts most prevalent (Meisel, Schwab, Prass, Meisel, & Dirnagl, 2005). Advanced age and stroke severity are known early predictors for poor patient outcome after stroke (Wartenberg et al., 2011), possibly due to elevated risk of infection, but the underlying mechanisms are unclear.
Randomized clinical trials evaluating prophylactic antibiotics in patients with acute stroke showed that this therapeutic approach did not lower the incidence of poststroke pneumonia (Kalra et al., 2015), and it was not associated with reduced mortality or improved functional outcomes (Xi et al., 2017). These trials suggest a clear need for alternative treatment approaches, and importantly, a better understanding of the underlying mechanisms of poststroke infections. Associative clinical studies speculate that risk factors for poststroke pneumonia include dysphasia, nasogastric tubing, catheters, mechanical ventilation and aspiration (Chapman, Morgan, Cadilhac, Purvis, & Andrew, 2018). In addition, the age-related decline in immune functions may further increase susceptibility of the elderly to poststroke infection (Crapser et al., 2016;Ritzel et al., 2018;Shaw, Goldstein, & Montgomery, 2013). While these factors may play an important role, our recent work demonstrated that poststroke infection may also originate from dissemination of gut commensal bacteria to peripheral tissues (Stanley et al., 2016). Indeed, >70% of the microorganisms detected in infected stroke patients of the studied cohort were common commensal bacteria that reside in the human intestinal tracts (e.g., Enterococcus spp., Escherichia coli and Morganella morganii) (Stanley et al., 2016). Unfortunately, a major caveat to these findings that demonstrate a vital brain-gut link in the setting of stroke is that the effect of age was not assessed and warrants further investigation. As such, in this study, we examined if advanced age can promote the translocation and dissemination of commensal bacteria to potentiate the development of poststroke infection.

| Advanced age is a risk factor for the development of infection in patients after acute stroke
A total of 633 patient records were obtained from the Monash Medical Centre, where 124 did not meet the inclusion criteria and were hence excluded, leaving a total of 509 within the study cohort. Patient cohort characteristics are summarized in Table 1

| Advanced age exacerbates lung infection in poststroke mice
In this study, we chose to use 12-to 15-month-old mice for modelling advanced age and denoted this group as "older." Significant mortality rate (44%) after stroke was recently reported with the use of animals >18 months old (Ritzel et al., 2018), as such we selected 12-to 15-month-old mice to represent our "older" group in order to avoid studying survival bias.
To assess the effect of advanced age on injury lesion development after the mild model of stroke, we performed preclinical magnetic resonance imaging (MRI) on our young and older cohort of animals at 24 hr following stroke onset. We observed similar oedema volumes after stroke between the two groups of mice ( Figure 1a), but mice in the older group showed enhanced neurological impairment compared with younger counterparts (Figure 1b). This is consistent with the well-reported clinical observations that elderly patients often present with greater neurological impairment and more debilitating outcomes after stroke. To understand how advanced age contributes to the development of poststroke infection, we quantified the amount of culturable bacteria from lung homogenates 24 hr after stroke. Older mice demonstrated a 100-fold increase in culturable bacteria after stroke compared with sham-operated controls and young counterparts, suggesting an impairment of antimicrobial defence in advanced age after stroke ( Figure 1c). Levels of culturable bacteria in the blood, liver, mesenteric lymph node (MLN) and spleen remained similar and largely undetectable for both older and young mice after stroke when compared to their respective sham-operated animals ( Figure S1).

| Older mice exhibit greater colonic permeability poststroke
Our previous study using a more severe model of transient stroke in young mice (7-10 weeks old) showed that poststroke infection is attributable to intestinal dysfunction and the subsequent systemic dissemination of host intestinal bacteria (Stanley et al., 2016). However, we are yet to understand whether age contributes to this phenomenon. To examine the effect of advanced age on gut permeability,

| Older mice show evidence of colonic barrier breakdown poststroke
Integrity of the colonic barrier encompasses many structural and physiochemical aspects that collaborate to regulate intestinal permeability and prevent bacterial translocation (Chelakkot, Ghim, & Ryu, 2018

| Reduced tight junction protein expression in older mice after stroke
Beyond the first line of defence, the colonic barrier is strengthened further by important multi-complex tight junctions that ensure strong scaffolding of intestinal epithelial and endothelial cells (Chelakkot et al., 2018). Alterations in the protein expression of tight junctions can lead to intestinal structure disruption and changes to paracellular and intercellular barrier permeability. Using immunofluorescence staining, we observed that the expression of key proteins of the tight junction complex, zonula occludens-1 (ZO-1), was reduced exclusively in the colon of older mice after stroke, and not their young counterparts ( Figure 4a). This pattern of reduced expression appeared specific to ZO-1 as it was not statistically differ-  Figure S2).

| Tumour necrosis factor-α facilitates the breakdown of colonic barriers poststroke
Given that older mice showed greater histological evidence of intestinal inflammation than their younger counterparts (Figure 2d), we hypothesized that the stroke-induced pro-inflammatory microenvironment promotes barrier breakdown in the colon of older mice.
Indeed, previous studies have reported that pro-inflammatory cytokines, specifically tumour necrosis factor (TNF-α), play a critical role in regulating tight junction proteins in a myosin light chain kinase (MLCK)-dependent manner, inducing intestinal permeability, pathology and inflammaging (Shen, 2012;Yu et al., 2010). In this study, expression of TNF-α and IL-10 was elevated exclusively in the colonic tissue of older, and not young mice 5 hr poststroke compared with sham-operated cohorts (Figure 5a,b): a time point that preceded any changes in tight junction expression of Cldn3, Cldn5, Ocldn and ZO-1 ( Figure S3). Additionally, we found no difference in the total number of CD45 + leucocytes, CD3 + T cells, CD11b + myeloid cells, Ly6G + / Ly6C − neutrophils or Ly6G − /Ly6C + monocytes in the colon, suggesting this early pro-inflammatory cytokine environment was not be driven by cellular immune changes ( Figure S4). By 24 hr poststroke, elevated TNF-α and IL-10 expression in the colon of older poststroke mice returned back to baseline sham levels ( Figure S5A Tumour necrosis factor-α signalling has been shown to participate in cerebral neuroinflammation after stroke (Hallenbeck, F I G U R E 1 Advanced age exacerbates neurological impairment and lung infection poststroke. The following assessments were performed 24 hr after mid-cerebral artery occlusion induction on young (7-10 weeks) and older (12-15 months) mice: (a) preclinical magnetic resonance imaging (MRI) to assess volume of brain oedema (n = 6/group). Oedema region indicated within yellow outline; (b) neurological assessment (n = 14-16/ group); (c) bacteriological analysis of lung homogenates to assess poststroke lung infection (n = 6-8/group). Data represent the mean ± SEM. Significance was determined by Mann-Whitney U test, and a p-value ≤0.05 was considered statistically significant: *p ≤ 0.05, **p ≤ 0.01

| D ISCUSS I ON
Aside from neurological damage, bacterial pneumonia is the most frequent severe complication, and the most common cause of death, in patients with stroke (Meisel et al., 2005;Shim & Wong, 2018).
Confirming previous clinical studies, we showed using a retrospective patient cohort that increasing age is an independent predictor for developing infection after stroke. Further to this, we utilized an villi, with functional impairment of isolated intestinal epithelial stem cells, suggesting a reduced capacity for tissue repair following injury (Moorefield et al., 2017).
Given that the gut barrier integrity and intestinal inflammatory balance are closely linked with gut microbiome, it is possible that age-dependent alterations in the microbiota further influence stroke outcomes. It is known that older individuals have a very different gut microbiota profile compared to healthy adults. Generally, the gut microbiome of elderly individuals is characterized by reduced bacterial diversity with a shift towards lower levels of beneficial populations such as bifidobacteria, and this is associated with increased frailty (Jackson et al., 2016;Nagpal et al., 2018). The mechanism of microbiota change with aging is not well understood, but it is thought that factors associated with cell senescence, immune changes, co-morbidities, intestinal physiology and lifestyle play a role. Several recent studies using high-throughput gene sequencing tools demonstrated that stroke induces robust changes to the intestinal mucosal microbiota, with an overall reduction in species diversity (Singh et al., 2016;Stanley et al., 2018). Furthermore, it is important to remember that brain-gut communication is bidirectional, with emerging evidence showing that disruption of microbial-host symbiosis prior to stroke also leads to altered infarct severity after cerebral ischaemia (Benakis et al., 2016).
A better understanding of age-associated alterations in mucosal microbiota composition, and their reciprocal impact on the host biological pathways, will assist in designing novel and targeted therapeutic approaches to reduce neurological damage and infectious complications after stroke.

| Mouse focal cerebral ischaemia model
The mid-cerebral artery occlusion (MCAO) model was performed as previously described (Nicholls, Wen, Hall, Hickey, & Wong, 2018;Stanley et al., 2016Stanley et al., , 2018. Young and older animals were given 20 min of MCAO followed by reperfusion to model a mild form of ischaemic stroke, resulting in <10% mortality in both groups to avoid survival bias in our study. Sham-operated animals underwent anaesthetic, neck incision and artery isolation only. All animals were individually housed after MCAO or sham surgery. See Data S1 for surgical details. To examine whether TNF-α can alter intestinal permeability in vivo, recombinant TNF-α (20 µg/kg) or saline as control was administered (i.p.) immediately following blood reperfusion to young stroke and sham-operated mice.

| Magnetic resonance imaging
Oedema volumes were measured using MRI 24 hr after MCAO in young and older mice. Measurements of oedema exclude the volume of the ventricle at the dorsal section of the infarct brain hemisphere.
See Data S1 for MRI scanning procedure and settings.

| Neurological assessment
At 24 hr after MCAO, neurological assessment was performed on young and older mice using an established six-point scoring system (Kim et al., 2014). See Data S1 for scoring parameters.

| Gut permeability assay
Mice were orally gavaged with 500 mg/kg of 4.4-kDa fluorescein-isothiocyanate-labelled dextran (FITC-dextran; Sigma) at 2 hr after MCAO. Animals were anesthetized with isoflurane at 1 hr (small intestine permeability) or 4 hr (colon permeability) after gavage, and a cardiac puncture performed to collect serum. Serum concentrations of FITC-dextran were determined relative to a standard curve (top standard at 1,250 µg/ml) at an excitation of 485 nm.

| Vascular permeability assay
At 24 hr after MCAO, animals were given an intravenous injection of 4 ml/kg 2% Evans blue (Sigma) in saline. Mice were then transcardially perfused with saline at 4 hr after Evans blue injection. The whole colon was homogenized in 3 ml of N, N-dimethylformamide (Sigma) and further incubated in N,N-dimethylformamide overnight at 55°C.
Supernatant containing Evans blue was collected by centrifugation at 500 g for 10 min, and concentrations of Evans blue determined relative to a standard curve (top standard at 1,000 µg/ml) at an excitation of 620 nm.

| Colon histological scoring
At 24 hr after MCAO, a section of the distal colon was collected for haematoxylin and eosin (H&E) staining. For each sample, two representative images of colonic crypts at 100× magnification were captured using the Leica DM LB widefield microscope and MC120 HD camera (Leica) for histological scoring in a blinded manner. Parameters for histology scoring were previously established (Shen et al., 2018) and detailed in Data S1. Total overall score indicates the degree of colonic pathology after MCAO or sham operation.

| Periodic acid-Schiff staining
At 24 hr after MCAO, a section of the distal colon was collected for Periodic acid-Schiff reagent (PAS) staining according to standard protocols. For each sample, two representative images of colonic crypts at 100× magnification were captured using the Leica DM LB widefield microscope and MC120 HD camera (Lecia). The number of goblet cells per colonic crypt was quantified and presented as number of cells per mm 2 .
Primer sequences are detailed in Table S3. Data were normalized to housekeeping gene 18S and analysed using the 2 (−ΔΔCt) method.
Gene expression was expressed as fold change relative to colon tissue from sham-operated animals.
Approximately 6-8 images were captured for each colon section for analysis and averaged. imagej (NIH) was used for image processing to quantify the area of ZO-1 respective to area of DAPI staining.
The LSR-Fortessa (BD Biosciences) and FlowJo (Tree Star) were used for acquisition/analysis.

| Ex-vivo assessment of TNF-α on colonic tight junction complexes
Colons of naïve older (12-15 months) male C57BL/6J mice were carefully excised, dissected into 1 cm cross sections, faecal matter gently removed and placed in carbonated Krebs buffer supplemented with 2 g/L d-glucose. Carbonated Krebs buffer is considered the most biorelevant buffer system for the simulation of intestinal conditions (Fadda, Merchant, Arafat, & Basit, 2009).
Colon cross sections were submerged into 1 ml of Krebs buffer (1 ml) and treated for a total of 1 hr with either 0 or 1 ng of recombinant mouse TNF-α (BioLegend). During treatment, the Krebs buffer and colon sections were constantly carbonated at the right pH to ensure tissue viability. Following treatment, tissues were processed for qRT-PCR to examine the role of TNF-α on tight junction complex expression.

| Statistics
Quantitative data for experimental mouse studies are presented as mean ± standard error of the mean (SEM). Statistical analyses were conducted using graphpad prism Software. Data sets were tested for normality using the Shapiro-Wilk normality test. Nonparametric data were analysed using the Mann-Whitney U test. Comparisons of multiple parametric data sets were analysed using one-way analysis of variance (ANOVA) with post hoc comparison with Holm-Sidak multiple testing correction. Single comparisons of parametric data sets were analysed using the Student t test. Adjusted p-value ≤0.05 was considered statistically significant.