Juvenile Mice Develop Infarct-Induced Neurodegeneration and Emerging Cognitive Decline in a Model of Pediatric Stroke

Abstract

BACKGROUND:

Over half of pediatric stroke survivors have permanent cognitive deficits, which can emerge late after stroke. Mechanisms of cognitive decline after pediatric stroke may differ from those in adults because children’s strokes occur while brain and immune development are still ongoing. We therefore aimed to develop a pediatric mouse model of infarct-induced delayed cognitive decline to define age-related differences in immune responses compared with adults.

METHODS:

Male and female C57BL/6J mice were randomized to stroke or sham surgery at 28 days old to model stroke in late childhood. We used permanent distal middle cerebral artery occlusion followed by 60 minutes of hypoxia to induce an ischemic cortical stroke. Juvenile mice underwent behavioral testing at 1 and 7 weeks after surgery using Barnes maze and Novel Object Recognition tests. We quantified stroke size, atrophy, and neuroinflammation at 3 days and 7 weeks after surgery in juvenile and adult mice using immunostaining.

RESULTS:

One week after surgery, juvenile stroke and sham mice performed comparably on cognitive testing. However, by 7 weeks after surgery, stroke mice of both sexes performed significantly worse on reversal learning with the Barnes maze. Histologically, juvenile mice had greater innate immune activation at sites of secondary neurodegeneration in the corpus callosum, corticospinal tract, and thalamus at 3 days after stroke, while adults had greater chronic innate immune activity at these sites 7 weeks after stroke.

CONCLUSIONS:

In a model of childhood ischemic stroke, juvenile mice of both sexes developed an emerging cognitive deficit analogous to that seen in pediatric stroke survivors. It was associated with chronic neuroinflammation in uninjured subcortical structures that undergo secondary neurodegeneration. Our results suggest that infarct-induced neurodegeneration occurs after stroke in juvenile mice, and that there are age-related divergent trajectories in the innate immune response to stroke at sites of secondary neurodegeneration.

The incidence of pediatric stroke is rising worldwide,^1–3 and up to 70% of survivors have permanent cognitive impairment that is characterized by prominent deficits in executive function, processing speed, and cognitive flexibility.^4–7 In children, impairment in these cognitive domains presents as learning disabilities, and stroke doubles the risk of developing learning disorders such as attention-deficit/hyperactivity disorder (ADHD) to 5.7% in survivors of childhood stroke.⁸ Furthermore, some survivors develop emerging or worsening cognitive symptoms years after their stroke, which is a well-recognized phenomenon but poorly understood.⁹ In 1 study, 70% of children with ADHD after stroke had declines in their reading scores over a 4-year period, whereas those with ADHD and no history of stroke had stable or improved reading scores.¹⁰ Learning and cognitive disabilities profoundly impair not only children’s quality of life but also their ability to transition to independent adulthood, with profound psychosocial and financial consequences.^11,12

Chronic cognitive deficits after pediatric stroke are associated with widespread structural changes in brain regions distant from the infarct, including both the cortex and white matter tracts of the uninjured hemisphere.^13–15 Adults may also have atrophy in uninjured brain regions due to secondary neurodegeneration after stroke.^16–18 However, some evidence suggests that chronic volume loss in uninjured brain regions, including the thalamus and corpus callosum, may be accentuated in children compared with adults.^19,20 Minimal autopsy tissue exists from pediatric stroke populations, limiting our ability to infer mechanisms underlying the widespread changes in brain growth that are observed clinically. Moreover, existing animal work in pediatric stroke has typically focused on processes occurring within the infarct itself at acute and subacute timepoints. Thus, the cellular and molecular mechanisms that drive infarct-induced neurodegeneration in children, leading to chronic cognitive disability, altered brain growth, and emerging cognitive decline after pediatric stroke, remain poorly understood.

Given the substantial burden imposed by chronic cognitive dysfunction after pediatric stroke and the paucity of knowledge about its underlying mechanisms, we therefore set out to develop a mouse model of delayed, infarct-induced cognitive decline after pediatric stroke and to determine whether this emerging cognitive decline was associated with chronic degenerative changes outside the infarct. We used 28-day-old mice (described subsequently as juvenile mice) to model stroke in late childhood.²¹ To identify the specific neurodegenerative mechanisms that drive late cognitive decline (rather than immediate, lesion-related deficits), it is essential that mice lack immediate cognitive impairment after stroke; thus, any cognitive impairment that emerges at chronic time points will be attributable to infarct-induced neurodegeneration. The cognitive domains impaired in the model should also align with those often affected in pediatric stroke survivors independent of infarct location, specifically processing speed, executive function, and cognitive flexibility.^4,5,7,22 Finally, it should induce secondary neurodegeneration in brain regions that are not acutely injured but that receive or send projections to the infarcted tissue, an essential requirement to interrogate the mechanisms that cause extensive changes in development throughout the entire brain after childhood stroke.^13,15

Both the innate and adaptive immune responses to stroke are implicated in poststroke cognitive decline in adult humans and mice.^23–27 However, it is unknown whether children or juvenile mice have chronic neuroinflammation after stroke, particularly in brain regions outside the infarct. In addition, children’s immune systems are immature, so responses may be distinct from those in adults.²⁸ Therefore, we also compared the acute (72 hours) and chronic (7 weeks) inflammatory responses to stroke between juveniles and adults in uninjured subcortical structures (thalamus, corpus callosum, and corticospinal tracts) that undergo secondary neurodegeneration.

Methods

Data Availability

Data related to this article are available from the corresponding author upon reasonable request.

Animals

C57BL/6J mice were purchased from Jackson Laboratories (Catalog 000664). Juvenile mice were postnatal day 28 and adult mice were aged 6 months at the time of surgery. Mice were group housed up to 5 mice per cage on a 12-hour light/dark cycle, with ad libitum access to food and water. All procedures were approved by the Stanford University Institutional Animal Care and Use Committee.

Experimental Procedures

Distal Middle Cerebral Artery Occlusion Stroke Model

Distal middle cerebral artery occlusion (dMCAO) stroke was performed as previously described.²⁹ The scalp was shaved and mice were anesthetized using isoflurane at 2.0 L/min. Temperature was maintained at 37 °C using a homeothermic heating system with a rectal probe to monitor core temperature (Stoelting Catalog 50300). The scalp was disinfected with chlorhexidine surgical scrub. Ophthalmic eye ointment was placed over the eyes (OptixCare Pet Eye Lube Plus, Aventix Animal Health). An incision was made over the skull and the right temporalis muscle was mobilized at the superior border. A surgical microdrill (Harvard Apparatus catalog 72-6065) was used to make a small craniotomy overlying the distal middle cerebral artery. The dura was dissected off the artery. For stroke surgeries, the middle cerebral artery was cauterized (Bovie DEL1 Change-A-Tip Deluxe High Temperature Cautery Pen). All steps were identical for sham surgery, except for cauterization of the artery. The temporalis was replaced and the surgical wound was closed with surgical glue (Meridian Surgi-Lock 2oc Instant Tissue Adhesive). All mice received subcutaneous injections of 0.5 mg/kg Buprenorphine ER (Wedgewood Veterinary Pharmacy) and Cefazolin 25 mg/kg (Cefazolin sodium salt, VWR catalog 89149-888) following surgery.

Distal Hypoxic Stroke

Distal hypoxic (DH) stroke was performed identically to dMCAO stroke, then followed by hypoxia as described.³⁰ After wound closure, the animal was allowed to recover for 5 minutes following the end of anesthesia, then placed in a hypoxia chamber (Coy vinyl anaerobic chamber; Coy Laboratory Products) containing 8% oxygen and 92% nitrogen for 60 minutes. Temperature in the hypoxia chamber was maintained at 37 °C. Both stroke and sham-operated mice underwent hypoxia identically.

Behavioral Testing

Testing was performed during the light phase without food or water restriction. Barnes maze³¹ and novel object recognition^31–33 tests were performed at 1 and 7 weeks after surgery (Figure S1A). For Barnes maze, we tested 2 protocols in independent cohorts of mice. In the Barnes maze reversal learning protocol, mice underwent 3 trials per day for 3 days of training; the release location was moved for each trial, and the location of the escape hole was moved each day to evaluate reversal learning and cognitive flexibility. For the Barnes maze spatial learning paradigm, mice underwent 3 trials per day for 3 days of training; the animal was released in the center of the maze for every trial, and the location of the escape hole remained constant throughout each time point. Automated behavior scoring software (BehaviorCloud, BehaviorCloud LLC) was used to measure time to locate the escape hole, distance traveled, and average speed. Novel object recognition was performed as previously described with a fixed 2.5-hour intertrial interval.^31,32,34 Open-field testing was performed as previously described³¹ to test for fatigue or weakness as confounding factors, and BehaviorCloud was used to evaluate center crossings, time in the center, average speed, and path length. Identical behavioral scoring criteria and analysis pipelines were applied across all experiments.

Tissue Preparation

All mice were terminally anesthetized with ketamine/xylazine, then perfused with ice-cold phosphate-buffered saline. After perfusion, brains were drop-fixed in 4% paraformaldehyde in phosphate buffer for 24 hours at 4 °C, followed by impregnation with 30% sucrose. Forty-micrometer-thick coronal brain sections were prepared using a freezing microtome (Microm HM430) and then stored at −20 °C in a cryoprotective solution containing 30% glycerin, 30% ethylene glycol, and 40% 0.5 mol/L sodium phosphate buffer.

Immunostaining

Immunostaining was performed as previously described.²⁶ We stained for reactive astrocytes (GFAP [glial fibrillary acidic protein], Agilent Dako Z0334, 1:10 000), microglia/macrophages (CD68: BIO-RAD MCA1957, 1:1000; Iba1: Fujifilm-Wako 019-19741, 1:1000; Iba1: Abcam ab5076, 1:500), neurons (NeuN: Sigma-Aldrich MAB377, 1:500), B lymphocytes (B220: BD Biosciences 553085, 1:1000), T lymphocytes (CD3ε: BD Biosciences 550277, 1:500), major histocompatibility complex class II-expressing cells (I-A/I-E: BD Pharmingen 553621, 1:500), and lipid droplets (Plin2: Proteintech 15294-1-AP, 1:500). Biotinylated secondaries (Vector Labs) included goat antihamster (BA-9100-1.5), goat antirabbit (BA-1000-1.5), and rabbit antirat (BA-4001-.5), used at 1:500-1:1000. Alexa Fluor–conjugated secondary antibodies (Invitrogen) were donkey antigoat 647 (A-21447) and donkey antirabbit 488 (A-21206), used at a concentration of 1:200.

Imaging

Images were acquired on a Keyence BZ-X700 series inverted bright-field microscope. Fluorescent imaging of Plin2 staining was performed on a Leica TCS-SPE confocal microscope. Quantification of staining in the infarct core and corpus callosum was done in 5 sections per mouse, and quantification of the thalamus and corticospinal tracts was done in 2 sections per mouse. Anatomic regions were located using the Allen Mouse Brain Atlas. Automated thresholding was performed using a custom ImageJ macro to reduce bias. For infarct volume analysis, every sixth section was stained using NeuN. The infarct core, ipsilateral hemisphere, and contralateral hemispheres were manually traced. Infarct volume was calculated using the Cavalieri method³⁵ and reported as a percentage of the ipsilateral hemisphere.

Statistical Analysis

Sample size was determined via power analysis using G*Power 3.1 based on a pilot experiment testing reversal learning on Barnes maze in juvenile mice. Power analysis indicated that a group size of n=10 per group would provide 80% power at α=0.05 to detect a 40% difference between stroke and sham groups in mean time to locate the escape hole on the final day of training. Normality testing was performed using GraphPad Prism 10 and nonparametric tests were used where appropriate. Comparisons of 2 groups at a single timepoint were done using Student t test (parametric) or Mann-Whitney U test (nonparametric). Comparisons of ≥3 groups were done with 1-way ANOVA, followed by Šidák correction for post hoc pairwise comparisons, or Kruskal-Wallis testing with Dunn correction for post hoc pairwise comparisons, as appropriate. Learning over the training period of Barnes maze was analyzed using a 2-way ANOVA with repeated measures. No animals that survived surgery were excluded from behavior testing or immunostaining analyses.

Randomization and Blinding

For all experiments, animals were randomly assigned to stroke or sham surgery in a 3:2 ratio by a custom-written Excel spreadsheet. Experimenters were blinded to surgery group (stroke or sham) throughout behavior testing and analysis, and throughout immunostaining analyses except where not possible (eg, in analysis of infarct volumes).

Results

We induced stroke by permanently occluding the distal middle cerebral artery because this model does not cause motor impairment. The immature brain may have more robust collateral blood flow than the adult brain,^36–38 so stroke models in juvenile mice might produce smaller infarcts. Therefore, we first quantified infarct volumes in 2 permanent distal occlusion models that produce delayed cognitive deficits in older mice: dMCAO²⁹ and DH stroke,³⁰ in which the dMCAO is followed by 60 minutes of hypoxia to produce larger, more consistent infarct volumes in adult C57BL/6J mice.

DH Stroke Induces a Reproducible, Cortical Stroke With Low Variability and Low Mortality in Juvenile Mice

We quantified infarct volume at 72 hours after stroke and chronic brain atrophy at 7 weeks after stroke (Figure 1B and 1C; Figure S1B and S1C). We chose these timepoints because 72 hours is the peak of the acute immune response to stroke,³⁹ and 7 weeks is when cognitive impairment emerges in the adult model of infarct-induced delayed cognitive decline.²⁶ dMCAO stroke in juvenile mice indeed produced small infarcts (Figure 1B) and failed to produce stroke in up to 20% of animals (6/28 animals over 3 experiments). The addition of 1 hour of hypoxia after stroke or sham surgery increased acute stroke size in both male and female juvenile mice to a volume comparable to that in adult (6-month-old) mice that underwent DH stroke (Figure 1B). Over time, DH stroke caused atrophy in the ipsilateral hemisphere, which was consistent between age groups and across sexes in juveniles (Figure 1C). We did not observe significant neuronal loss in the ipsilateral hippocampus in juveniles at 7 days after DH stroke (Figure 1D and 1E), consistent with previous work showing that hypoxia does not cause neuronal loss in either sham-operated adult mice or outside the infarct in stroke-operated adults.³⁰ Survival from stroke surgery did not differ significantly across sex or age (juvenile females: 20/24, 83%; juvenile males: 73/96, 76%; adult males: 16/20, 80%). Survival during sham surgeries was higher than during stroke surgeries (juvenile females: 12/13, 92%; juvenile males: 44/51, 84%; adult males: 8/8, 100%). All animals that survived the initial surgery were analyzed in subsequent experiments.

Figure 1.

Juvenile distal hypoxic (DH) stroke. A, 2,3,5-triphenyltetrazolium chloride (TTC) staining of juvenile DH stroke at 72 hours after surgery. B, Acutely (72 hours after surgery), juvenile distal middle cerebral artery occlusion (dMCAO) stroke produces small infarcts, while juvenile DH stroke produces larger infarcts comparable in volume to adult animals. There is no significant difference in acute infarct volume across DH stroke groups (F[2, 13]=0.9507, P=0.4118, n=4–7/group, 1-way ANOVA). There is no difference in infarct volume between juvenile male DH stroke and juvenile female DH stroke (P_adj=0.7986, 1-way ANOVA with Šidák correction for multiple comparisons). C, At 7 weeks after surgery, DH stroke induces volume loss in the ipsilateral hemisphere of both juveniles and adults. There is no significant difference in chronic brain atrophy across groups (F[2, 20]=2.483, P=0.1088, 1-way ANOVA, n=6–9/group) or between juvenile male and female mice (P_adj=0.0937, 1-way ANOVA with Šidák correction for multiple comparisons). D, Representative Fluoro-Jade C staining for cell death from the infarct core (top) and ipsilateral hippocampus (bottom) from juvenile male mice at 7 days after stroke. E, Quantification of Fluoro-Jade-positive cells in the infarct and ipsilateral hippocampus (n=7 mice, 2 images per mouse in each location, nested t test). ***P<0.001. Error bars: mean±SEM.

DH Stroke in Juvenile Mice Causes a Delayed, Infarct-Induced Impairment in Reversal Learning

To test for delayed cognitive decline in DH stroke in juvenile mice, we used the Barnes maze and novel object recognition task at both an acute time point (1 week after stroke) and a chronic time point (7 weeks after stroke; Figure S1A). We first tested reversal learning as a correlate of cognitive flexibility,⁴⁰ which is often impaired in pediatric stroke survivors^7,22,41 (Figure 2A), then, in an independent cohort, spatial memory (Figure 2B and 2C), because it is impaired after dMCAO and DH stroke in adult mice.^26,33

Figure 2.

Juvenile distal hypoxic (DH) stroke induces a delayed, selective impairment in reversal learning. A, Left: Schematic of the reversal learning paradigm. The release location (where mice are placed to start the maze task) changes on each trial, and the location of the escape hole changes each day. Right: Quantification of the mean time to locate the escape hole. Juvenile stroke and sham mice perform comparably 1 week after stroke (F[1, 65]=0.006885, P=0.9341, 2-way ANOVA with repeated measures, n=29 sham, 38 stroke from 3 independent cohorts). However, at 7 weeks after stroke, the stroke mice have a deficit in reversal learning and perform significantly worse than sham mice (F[1, 65]=21.32, P<0.0001, n=29 sham, 38 stroke, 2-way ANOVA with repeated measures). B, Left: Schematic of the spatial memory paradigm. Both the release location and the escape hole location are fixed across all trials and all days of training. Right: Juvenile stroke and sham mice perform comparably at both 1 and 7 weeks after stroke on the spatial memory task (Week 1: F[1, 27]=0.7983, P=0.3795; Week 7: F[1, 27]=0.04952, P=0.8256, n=14 sham, 15 stroke, 2-way ANOVA with repeated measures). C, Left: Schematic of the novel object recognition task. Right: Juvenile stroke and sham mice both exhibit a preference for the novel object, indicated by a discrimination index (DI) >0 at 1 and 7 weeks after stroke, n=14 to 15/group, Wilcoxon rank-sum test. There is no difference between stroke and sham groups at either time point (n=14–15/group, Mann-Whitney U test). *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Error bars: mean±SEM.

One week after surgery, there was no difference between juvenile male sham-operated and stroke-operated mice in performance on the reversal paradigm. However, by 7 weeks after surgery, the stroke-operated mice developed a deficit in reversal learning. Although both groups performed similarly on the first day of testing, the sham-operated mice continued to improve throughout the following days, while the stroke-operated mice performed progressively worse through the testing period (F[1, 65]=21.32, P<0.0001, 2-way ANOVA). By the final day of testing, stroke-operated mice required 64% more time than sham-operated mice to locate the escape hole (sham group: 28±1.8 seconds; stroke group: 46±2.4 seconds; Figure 2A). Because infarct-induced cognitive decline has not previously been reported in models of pediatric stroke, we confirmed that this finding was reproducible by 2 separate experimenters in each of 3 independent cohorts of male mice with group sizes of 7 to 15 per group (total 29 shams, 38 strokes). A separate cohort of juvenile mice underwent open-field testing to evaluate fatigue or weakness as confounding factors. There was no difference in average speed or total distance traveled on open-field testing at 1 or 14 weeks after stroke in juvenile mice (n=14 sham, n=15 stroke; Figure S2). Thus, the deficit at 7 weeks after stroke does not reflect fatigue or motor dysfunction. Juvenile female mice also developed infarct-induced impairment in reversal learning at 7 weeks after surgery (F[1, 23]=4.796, P=0.0389, n=12 shams, 13 strokes, 2-way ANOVA; Figure S3).

We next tested whether spatial memory is affected in this model by adjusting the Barnes maze paradigm to test spatial learning (Figure 2B). There was no difference between juvenile male stroke-operated mice and sham-operated mice at either 1 or 7 weeks after surgery on the Barnes maze spatial paradigm. The novel object recognition task also tests short-term memory. Congruent with the results on the spatial learning version of the Barnes maze, we also did not observe a difference between stroke- and sham-operated juvenile male mice on the novel object recognition task (Figure 2C). Both stroke- and sham-operated groups exhibited a preference for the novel object at both time points (median discrimination index significantly different from 0).

There Is Greater Acute Myeloid Cell Inflammation at Sites of Secondary Neurodegeneration in Juvenile Stroke Than Adult Stroke

We next compared the acute (72 hours) and chronic (7 weeks) inflammatory responses to stroke between juveniles and adults in uninjured subcortical structures (thalamus, corpus callosum, and corticospinal tracts) to evaluate age-dependent immune changes. At 72 hours after stroke, there is acute neuroinflammation within the infarct and peri-infarct cortex characterized by the presence of reactive astrocytes and activated myeloid cells. There was no difference across age or sex in CD68 staining for activated myeloid cells or GFAP staining for reactive astrocytes within the infarct and peri-infarct cortex (Figure 4A and 4B). Antigen presentation by microglia and macrophages may contribute to the recruitment of adaptive immune cells and chronic neuroinflammation after stroke,⁴² so we quantified the number of MHCII-expressing cells. MHCII staining for antigen-presenting cells in the infarct was comparable across age and sex (Figure 4C). Handling of lipid debris from dying cells is another function of microglia/macrophages after stroke, and emerging evidence links lipid droplet accumulation in microglia to cognitive decline in neurological disease.^43,44 We therefore quantified lipid droplet accumulation in macrophages/microglia by costaining for the microglia/macrophage marker Iba1 and the lipid droplet marker Plin2. At 72 hours after stroke, over 80% of Iba1+ cells in the infarct core contained Plin2+ lipid droplets, with similar Plin2+ immunoreactivity across age and sex (Figure 4D and 4E).

By 72 hours after stroke in adult mice, Wallerian degeneration leads to innate immune activation along axonal tracts connected to the infarct.⁴⁵ We therefore next evaluated brain regions that are highly connected to the somatosensory cortex and thus expected to undergo secondary degeneration after the infarct, such as the corpus callosum, thalamus, and corticospinal tracts (Figure 3A through 3C). Interestingly, juvenile stroke mice had significantly more CD68 immunoreactivity than adult mice in the contralateral corpus callosum, corticospinal tract, and thalamus (Figure 3). Iba1 staining showed a similar increase in expression in the same regions in juveniles compared with adults (Figure S5). MHCII+ cells were also found in the corpus callosum and the thalamus but did not differ in number between sexes or age groups (Figure S6). We did not observe Plin2 immunoreactivity in the corpus callosum in any of the groups. The increased microglial response in connected regions was not mirrored in astrocytic GFAP expression, which did not differ between ages (Figure S7).

Figure 3.

There is greater acute myeloid cell activation at sites of secondary neurodegeneration in juvenile mice at 72 hours after distal hypoxic (DH) stroke. A, Left: Quantification of CD68 staining in the ipsilateral thalamus (denoted by the blue box in the middle). One-way ANOVA: (F[2, 14]=9.429, P=0.0025). Šidák-corrected comparisons showed that juvenile males had significantly greater CD68 immunoreactivity than adults (P_adj=0.0426), with no significant difference between juvenile male and female mice (P_adj=0.1684). B, Quantification of CD68 staining in the contralateral corpus callosum (denoted by the blue box in the middle). One-way ANOVA: (F[2, 14]=14.04, P=0.0005). Šidák-corrected comparisons showed that juvenile males had significantly greater immunoreactivity than adults (P_adj=0.0015), with no significant difference between juvenile males and females (P_adj=0.7534). C, Quantification of CD68 staining in the corticospinal tract (denoted by the blue box in the middle). One-way ANOVA: (F[2, 14]=4.997, P=0.0230). Šidák-corrected comparisons showed that there was no significant difference between male and female juveniles, or between juvenile and adult males, in CD68 immunoreactivity in the ipsilateral corticospinal tract (P_adj=0.2802). Scale bar 100 µm. Quantification denotes mean±SEM. Statistics: 1-way ANOVA with Šidák correction for multiple comparisons, n=6 juvenile males; 5 juvenile females; 6 adult males. *P<0.05, **P<0.01.

DH Stroke in Juvenile Mice Exhibits Attenuated Chronic Myeloid Cell Neuroinflammation in Regions of Secondary Neurodegeneration Compared With Adult Stroke

Given the unexpected finding that stroke elicited greater chronic myeloid cell activation in the connected brain regions of juvenile mice than adult mice, we next asked whether this resulted in differences in chronic myeloid inflammation between juveniles and adults at 7 weeks after surgery in the same regions shown in Figure 3.

Compared with sham mice, both juvenile and adult stroke-operated mice had increased myeloid cell activation in the corpus callosum, corticospinal tract, and thalamus at 7 weeks after surgery (Figure 4A through 4C). However, when comparing juvenile to adult mice, we found the reverse of what we observed acutely after stroke. Specifically, in chronic stroke, juveniles exhibited significantly less CD68 staining in the thalamus and corticospinal tract, and a trend toward less CD68 staining in the corpus callosum (Figure 4). This was paralleled by less MHCII+ staining in both the thalamus and corpus callosum at chronic time points in the juvenile mice (Figure S8). In contrast to the high percentage of Plin2-positive microglia/macrophages in the infarct at 72 hours, we observed minimal lipid droplet accumulation in the infarct at this chronic time point (Figure S9).

Figure 4.

Both juvenile and adult male mice have chronic myeloid cell immunoreactivity at sites of secondary neurodegeneration, but it is attenuated in juveniles. A, Left: Quantification of CD68 immunostaining in the ipsilateral thalamus. One-way ANOVA: (F[3, 24]=139.6, P<0.0001). Šidák-corrected comparisons showed that there was significantly less CD68 immunoreactivity in the ipsilateral thalamus of juvenile stroke mice compared with adult stroke mice (P_adj <0.0001). Both stroke groups had significantly greater CD68 immunoreactivity than shams (juvenile stroke vs juvenile sham: P_adj=0.0260; adult stroke vs adult sham: P_adj<0.0001). Right: Representative images of CD68 immunostaining in the ipsilateral thalamus in the region denoted by the blue box in the middle. B, Left: Quantification of CD68 immunoreactivity in the contralateral corpus callosum. One-way ANOVA: (F[3, 26]=17.64, P<0.0001). There was a nonsignificant trend toward diminished CD68 immunoreactivity in the contralateral corpus callosum of juveniles compared with adults (P_adj=0.0908, Šidák correction for multiple comparisons). Both stroke groups had significantly greater CD68 immunoreactivity than shams (juvenile stroke vs juvenile sham: P_adj=0.0166; adult stroke vs adult sham: P_adj<0.0001). Right: Representative immunostaining of the contralateral corpus callosum in the region denoted by the blue box in the middle. C, Quantification of CD68 immunoreactivity in the ipsilateral corticospinal tract. One-way ANOVA: (F[3, 22]=32.87, P<0.0001). There was significantly less CD68 immunoreactivity in the ipsilateral corticospinal tract of juvenile stroke mice compared with adult stroke mice (P_adj=0.0006, Šidák correction for multiple comparisons). Both stroke groups had significantly greater immunoreactivity than shams (juvenile stroke vs juvenile sham: P_adj=0.0003; adult stroke vs adult sham: P_adj<0.0001). Right: Representative CD68 immunostaining of the ipsilateral corticospinal tract in the region denoted by the blue box in the middle. Scale bar 100 µm. Statistics, 1-way ANOVA with Šidák correction for multiple comparisons, n=6 to 10 males/group. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001. Error bars: mean ± SEM.

DH Stroke in Juvenile Mice Causes Comparable Chronic Lymphocytic Neuroinflammation in the Infarct

Because chronic adaptive neuroinflammation has previously been implicated in infarct-induced cognitive decline in adults,^23,26,46 we next determined whether juvenile stroke causes chronic lymphocytic neuroinflammation. Juvenile DH stroke caused chronic B- and T-lymphocyte neuroinflammation in the infarct scar that persisted for at least 7 weeks after stroke and was not different between juvenile stroke mice and adult stroke mice (Figure 5).

Figure 5.

Male juvenile and adult mice have comparable chronic lymphocytic neuroinflammation after stroke. A, Left: Quantification of B220 immunostaining for B lymphocytes in the infarct scar at 7 weeks after stroke. One-way ANOVA: (F[3, 26]=9.510, P=0.0002). Both juveniles and adults have similar chronic B lymphocyte neuroinflammation in the infarct scar at 7 weeks after stroke (P_adj=0.2358). While adult strokes had significantly greater B220 immunoreactivity than adult shams (P_adj=0.0004), juvenile shams were not significantly different from juvenile strokes (P_adj=0.0694). Right: Representative immunostaining for B220 in the infarct scar. B, Left: Quantification of CD3 immunostaining for T lymphocytes shows chronic T-lymphocyte neuroinflammation in the infarct at 8 weeks after stroke in juvenile and adult mice. One-way ANOVA: (F[3, 26]=13.54; P<0.0001). Juvenile stroke and adult stroke mice have similar CD3 immunoreactivity (P_adj=0.4294). Right: Representative immunostaining for CD3 in the infarct scar. Scale bar 100 µm. Statistics: 1-way ANOVA with Šidák correction for multiple comparisons, n=6 to 10 males/group. **P<0.01, ***P<0.001. Error bars denote mean±SEM.

Discussion

Here, we present a novel model of infarct-induced cognitive decline after cortical stroke in juvenile mice. The robust cognitive impairment we observed is accompanied by acute and chronic neuroinflammation both within the infarct and in uninjured subcortical regions that send or receive projections from the infarct and subsequently undergo secondary neurodegeneration. Intriguingly, while juveniles and adults have comparable acute myeloid and astroglial inflammation within the infarct, juveniles have markedly greater myeloid cell activation in uninjured subcortical structures including the thalamus, corpus callosum, and internal capsule. In contrast, at chronic time points, the opposite was true. Adults exhibited greater myeloid neuroinflammation in subcortical structures at 8 weeks after stroke. However, there was no difference in lymphocytic infiltration of the stroke scar between juveniles and adults.

Our finding that the developing brain is susceptible to infarct-induced cognitive decline is a major shift from the historical conceptualization of pediatric stroke as a focal, static brain injury. It also suggests that cognitive testing at earlier time points in juvenile mice may miss this delayed emergence of cognitive decline.⁴⁷ Our results therefore highlight the need for models of chronic cognitive dysfunction in a disease where the majority of survivors are expected to live for decades after stroke. Crucially, our model recapitulates deficits in mice that are similar to some of the cognitive domains that are often affected after stroke in children independent of infarct location—namely, cognitive flexibility and processing speed.^{4,5,9,48–50} While additional neuropsychological deficits (such as visuospatial processing) have been reported in some cohorts, a common unifying theme across studies is impairment in tasks requiring complex or higher order cognitive skills.^5,51 Notably, commonly used rodent behavioral assays designed to specifically test hippocampal function may be less sensitive for detecting deficits in executive function, processing speed, and other tasks requiring complex cognitive skills.

Infarct-induced autoimmunity is strongly implicated in driving delayed cognitive decline in adults.^25,26 We show that juvenile stroke induces chronic innate and adaptive neuroinflammation, both within the infarct and in connected white matter tracts and deep gray nuclei outside the infarct, which has not previously been reported in juvenile models. Because we did not observe significant chronic neuroinflammation in the juvenile shams, which also underwent hypoxia, the chronic inflammation seen in the juvenile stroke mice is likely a consequence of the infarct, not of the period of hypoxia used in the DH stroke model (although we cannot completely exclude that the combination of hypoxia and infarct could interact differently with the immature immune system than with the mature immune system). We are the first, to our knowledge, to demonstrate that chronic neuroinflammation remote from the infarct persists in juveniles for months after stroke and may be associated with impairment in cognitive flexibility. Additionally, our results suggest a potential causal mechanism by which focal injury can drive the wider disruption of both gray and white matter structures that have been reported clinically, ultimately impairing cognitive function.^13,15,19,20 Determining whether chronic inflammation plays a causal role in late-onset cognitive decline after pediatric stroke will be a vital focus for future research.

Finally, we observed chronic lymphocytic inflammation in the infarct that was similar between juveniles and adults. B lymphocyte-mediated neuroinflammation mediates delayed, infarct-induced neurodegeneration in adult mice, and depleting B lymphocytes prevents cognitive decline.²⁶ This identifies a possible mechanism with potential for therapeutic intervention, as B lymphocyte-depleting therapies are already widely used in pediatric demyelinating diseases,^52,53 with emerging evidence from Phase 2 trials suggesting cognitive benefit.⁵⁴

Our study identified crucial differences between the juvenile and adult brain after stroke. Juvenile mice developed a selective deficit in reversal learning, a correlate of cognitive flexibility. This is remarkably analogous to the cognitive deficits seen in survivors of childhood stroke, who often have profound impairments in executive function, cognitive flexibility, and processing speed that are independent of infarct location.^4,5,49 However, in juvenile mice, short-term memory (as tested by the novel object recognition and spatial Barnes maze) was relatively spared, while our group has previously shown that adult mice develop delayed, infarct-induced deficits in hippocampal long-term potentiation together with impairment on novel object recognition²⁶ and a version of the Barnes maze similar to the spatial paradigm used here.³³ Together, this suggests that there may be important differences in the neural basis of infarct-induced cognitive decline between the developing and adult brain and highlights the translational validity of our model.

Our findings that the juvenile and adult myeloid response to stroke diverges sharply over time add meaningful depth to the existing literature on age-dependent immune responses to stroke. While microglia seem to play an overall protective role in the acute phase after both neonatal^55,56 and adult stroke,^57,58 their specific functions, such as lipid handling, phagocytic capacity, production of reactive oxidative species and cytokines, and interactions with vascular and immune cells differ and may even diverge depending on age at stroke.^59–61 Much of this literature compares neonatal to adult stroke, or younger adults to aged adults. However, limited evidence suggests that the immune and vascular response to ischemic injury in childhood may also differ markedly from both the neonatal and adult periods.^62–64 In addition to changes in microglial number between juvenile and adult mice, there are also likely age-related changes in microglial activity and function both before and after stroke.⁶⁵

The data we present here show that age-related differences in neuroinflammation at sites of secondary neurodegeneration persist into the chronic phase of stroke. A major implication of this work is that chronic neuroinflammation may modulate long-term outcomes differently after pediatric stroke than after adult stroke, highlighting the importance of age-specific models of stroke. We observed little difference in inflammation within the infarct itself, but striking age-related differences in the trajectory of innate immunity at sites of secondary neurodegeneration. The acute myeloid cell inflammation outside the infarct was more pronounced in juveniles compared with adults. Intriguingly, this relationship reversed at chronic time points, when adult mice showed greater persistent innate neuroimmunity in these sites of secondary neurodegeneration. Few studies have compared inflammatory responses at sites of secondary neurodegeneration between younger adults and aged adults, and they report conflicting results. For example, Kluge et al⁶⁶ did not find age-related differences in microglial or astrocyte protein expression profiles in the thalamus at 28 days after cortical stroke. In contrast, Kim et al⁶⁷ reported that young adult mice had substantially greater persistent Iba1 staining in the ipsilateral thalamus at 14 days after cortical stroke than aged mice, despite no age-related differences in microglial or astrocytic activity within the infarcted tissue.

This difference in chronic myeloid inflammation at sites of secondary neurodegeneration is likely relevant to long-term outcomes and opens exciting new avenues to investigate potential causal roles of chronic neuroinflammation in pediatric infarct-induced cognitive decline. Microglia participate in synaptic pruning during both development and disease, including stroke.^68,69 Chronically activated microglia in regions such as the thalamus could engage in pathological synaptic pruning, potentially leading to the aberrant thalamocortical connectivity that is associated with cognitive impairment in pediatric stroke survivors.⁷⁰ Microglia also play multifaceted roles in developmental myelination and remyelination after injury, so chronic microglial inflammation in white matter tracts distant from the infarct could alter myelination patterns in pediatric stroke survivors.^71–73 Over time, this could lead to brain-wide alterations in white matter integrity observed chronically in survivors of pediatric stroke, which are associated with altered functional connectivity and cognitive impairment.^70,74–76 Finally, cytokine production⁷⁷ or antigen presentation by activated myeloid cells could also recruit or sustain the adaptive immune response,⁷⁸ which has been linked to cognitive decline after adult stroke in both animal and human studies.^23–26,33

Together, these processes could drive atrophy or disrupted development of structures such as the thalamus and corpus callosum, which are critical for cognitive flexibility, executive function, and reversal learning.^79–82 Atrophy of both the thalamus and corpus callosum has been observed in pediatric stroke survivors,^19,20,83,84 and worse cognitive outcomes are associated with altered functional thalamocortical connectivity.⁷⁰ Thus, these are plausible structural correlates of the observed deficit in reversal learning.

We also observed that microglia/macrophages within the infarct at 72 hours had prominent lipid droplet accumulation, while those in the corpus callosum had minimal lipid droplet accumulation. This suggests that even the same cell type may have important differences in function depending on location. Thus, our findings emphasize the importance of investigating age-dependent immune responses not only within the infarct but also at sites of secondary neurodegeneration, which is increasingly implicated in cognitive decline after adult stroke.^70,85,86

The divergent trajectory of inflammation between juvenile and adult mice also highlights the importance of the time course of inflammation, which may play differential roles at different timepoints. Further investigation of intermediate time points between the acute (72 hours) and chronic (7 weeks) timepoints we report here could illuminate the mechanisms that resolve inflammation in juvenile mice. The persistent myeloid cell activation we observe in adults relative to juvenile mice in the current study could suggest a failure of microglia and macrophages to terminate this acute, proinflammatory response and transition to an anti-inflammatory transcriptional profile in adult mice. The chronic MHCII staining we observed in adults is consistent with this hypothesis, as it implies that antigen presentation may be ongoing in adult mice late after stroke. This divergent trajectory in innate immunity at sites of secondary neurodegeneration may contribute to age-related differences in outcome after stroke between children and adults. Future studies with single-cell resolution using RNA sequencing and flow cytometry will provide additional insight into whether there are also important transcriptional, functional, or phenotypic differences between the myeloid cells that participate in the innate response to stroke in the pediatric and adult brain.

Interestingly, we did not observe significant sex-based differences in infarct volume, brain atrophy, cognitive outcome, or acute neuroinflammation in the juvenile mice. Identifying the mechanisms of sex-based differences in stroke risk^87–89 and outcomes⁹⁰ reported in clinical studies of pediatric stroke is an important area for ongoing work.

Cognitive impairment is common after pediatric stroke, affecting over half of survivors. Stroke roughly doubles the risk of developing ADHD and other learning disabilities, which has long-lasting implications for children’s health, quality of life, and transition to independence as adults.⁹¹ As in adult poststroke cognitive impairment, some children seem to be more vulnerable,⁹² and a subset of children may worsen over time.¹⁰ In addition to a limited understanding of which children are more vulnerable, the cellular mechanisms that cause children’s cognitive deficits after stroke remain unknown. Because of the challenges in conducting clinical research into children’s long-term cognitive outcomes and the lack of autopsy tissue, identifying these mechanisms requires development of translational models that complement our clinical studies. Here, for the first time, we report that focal cortical stroke induces chronic, widespread neuroinflammation in the juvenile brain, and that this is associated with infarct-induced cognitive decline. This model will support investigation into the mechanisms that drive chronic cognitive dysfunction after pediatric stroke. Identifying both biomarkers and therapeutic targets in the subacute and chronic periods after pediatric stroke is of paramount importance, since the majority of survivors have permanent neurological disability,⁹³ conferring a sizeable burden of lifelong morbidity without existing disease-modifying treatments.

ARTICLE INFORMATION

Acknowledgments

The authors thank Heather Fullerton for serving as an external advisor to Dr Mayne, and Karen Bradshaw and Li Zhu for assistance with sample collection and processing.

Author Contributions

Dr Mayne, K. Vanden, and M.E. Hefferon performed the experiments and analyzed the data. Drs Mayne, Zera, and Buckwalter designed the study and interpreted the results. Drs Mayne and Buckwalter drafted the article. All authors reviewed and approved the submitted version of the article.

Disclosures

Dr Mayne reports grant funding from the American Heart Association (AHA). Dr Zera reports grant funding from the AHA. Dr Buckwalter reports compensation from Genentech USA, Inc; F. Hoffmann-La Roche AG; and EMD Serono for consultant services on topics unrelated to this work, and grant funding from the Paul G. Allen Family Foundation. The other authors report no conflicts.

Supplemental Material

Figures S1–S9

ARRIVE 2.0 Checklist

Supplementary Material

str-57-2467-s001.pdf ^{(1.1MB, pdf)}

Funding Statement

This work was supported by a National Institutes of Health (NIH) National Institute of Neurological Disorders and Stroke–Child Neurology Career Development Program K12 award (K12 NS098482; Dr Mayne), an American Heart Association (AHA) Career Development Award to Dr Mayne (25CDA1451092), an AHA Career Development Award to Dr Zera (25CDA1435865), the Stanford Maternal and Child Health Research Institute (Dr Mayne), the AHA/Allen Frontiers Group Brain Health Award (Dr Buckwalter), a Leducq Transatlantic Network of Excellence award (Dr Buckwalter), and an Alzheimer’s Association Research Fellowship (Dr Zera).