Recognition Memory Impairments After Subcortical White Matter Stroke in Mice
Background and Purpose—Small subcortical white matter infarcts are a common stroke subtype often associated with cognitive deficits. The lack of relevant models confined to white matter has limited the investigation of its pathophysiology. Here, we examine tissue and functional outcome after an ischemic lesion within corpus callosum in wild-type (WT) mice and in mice null for a gene, NOTCH3, linked to white matter ischemic injury in patients.
Methods—WT and NOTCH3 knockout mice were subjected to stereotactic microinjections of the potent vasoconstrictor endothelin-1 at the level of periventricular white matter to induce a focal ischemic lesion. Infarct location was confirmed by MRI, and brains were examined for lesion size and histology; behavioral deficits were assessed ≤1 month in WT mice.
Results—Ischemic damage featured an early cerebral blood flow deficit, blood–brain barrier opening, and a lesion largely confined to white matter. At later stages, myelin and axonal degeneration and microglial/macrophage infiltration were found. WT mice displayed prolonged cognitive deficit when tested using a novel object recognition task. NOTCH3 mutants showed larger infarcts and greater cognitive deficit at 7 days post stroke.
Conclusions—Taken together, these data show the usefulness of microinjections of endothelin-1 into periventricular white matter to study focal infarcts and cognitive deficit in WT mice. In short-term studies, stroke outcome was worse in NOTCH3 null mice, consistent with the notion that the lack of the NOTCH3 receptor affects white matter stroke susceptibility.
The vast majority of experimental stroke research has focused on animal models of large territorial infarction involving both cortical and subcortical grey and white matter.1 However, 20% to 25% of strokes in humans are small, lacunar-like, involve subcortical white matter tracts,2 are often recurrent, and may lead to cognitive decline, subcortical dementia, and major disability.3 Moreover, white matter ischemic lesions are a main pathological manifestation of small vessel diseases, a subset of cerebrovascular alterations leading to stroke and cognitive decline.4 Only a few experimental models investigating small white matter strokes have been reported.5
Recently, Carmichael and collaborators6 developed a model of selective white matter stroke by injecting the potent vasoconstrictor endothelin-1 into corpus callosum to create a small area of myelin and axonal degeneration. Although reproducibility has been a problem with this model, it generates a lesion confined to white matter and therefore offer advantages to other white matter injury models that in addition cause diffuse brain injury after global cerebral hypoperfusion.7 After adapting, improving, and extending this new experimental approach, we tested whether subcortical white matter ischemic injury is associated with cognitive impairment in wild-type (WT) mice. Furthermore, we tested whether a NOTCH3 gene deletion negatively affects tissue and functional outcome after selective white matter injury. NOTCH3 mutations are the most common cause of inherited strokes and vascular dementia in young and middle-aged adults.8 We previously showed that NOTCH3 knockout mice seem especially vulnerable to ischemic injury after middle cerebral artery occlusion, probably because of vascular dysfunction and reduced collateral blood flow.9 Here, we extended these findings to white matter susceptibility and provided a novel experimental approach to further dissect the relationship between genotype and ischemic phenotype in NOTCH3 mutants.
Materials and Methods
Experiments were conducted according to protocols approved by the Animal Research Committee of Massachusetts General Hospital and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (see online-only Data Supplement).
NOTCH3 knockout male mice (N3KO; 23–28 g; n=8; genetically engineered as previously described)9 and sex-, age-, and weight-matched WT (n=34 for method development; n=89 for hypothesis testing) mice (C57BL/J, Charles River Laboratories; background strain) were housed 4 per cage and maintained on a 12/12-hour light/dark cycle and fed ad libitum. Only a limited number of mutants were available for this study. Mice were randomly allocated and coded with tail marks to perform the analysis in a blinded fashion. The number of mice needed for behavioral and infarct size assessments was based on power calculation assuming a group difference of 30% and an SD of 30% to 35%.
Endothelin-1 Stroke Model
Subcortical white matter stroke was produced using a recently published method,6 with slight modifications. Isoflurane anesthetized mice (4% for induction, 1%–1.5% for maintenance, in a 70:30 N2O:O2 mixture) were placed on a stereotactic frame (Stoelting), the head secured with blunted earbars and the body temperature controlled (37°C) using a feedback-regulated homeothermic blanket (FHC). Eye ointment was applied to prevent corneal dryness. After a midline scalp incision, a small (1×1 mm) burr hole was drilled under constant saline cooling, carefully keeping the dura intact, and 2×100-nL microinjections of vehicle (sterile saline, control group) or endothelin-1 (0.3 mg/mL, American Peptide, cat. 88-1-10, Lot. Z05068T1) were performed using a glass micropipette (tip≈25 μm) connected to a pressure system (Picospritzer, General Valve). To target the periventricular white matter at the level of corpus callosum, we adopted the following stereotactic coordinates: antero-posterior +1.00, +0.60 mm; medio-lateral −0.20 mm; dorso-ventral −2.20 mm; and 36° angle. After each injection, the pipette was left in place for an additional 5 minutes to avoid backflow. At the end of the procedure, mice were placed in a 28°C incubator for 2 hours before returning to the home cage.
Assessment of Cerebral Blood Flow and Blood–Brain Barrier Permeability
Regional cerebral blood flow (n=5 control, 5 stroke) was analyzed using the [14C]-iodoantipyrine ([14C]-IAP) method as previously described.10 Blood–brain barrier (n=5 control, 5 stroke) permeability was assessed using the Evans blue technique.11
Magnetic Resonance Imaging
Mice (n=5) were imaged 2 days after stroke. A change in magnetization transfer ratio was used to assess white matter damage because it highly correlates with myelin content.12
Mice were tested for sensorimotor and cognitive performances at 7 (n=18 control, 27 stroke WT, and 8 stroke N3KO) and 28 days (n=10 control, 9 stroke) after either vehicle or endothelin-1 injection. Cylinder (forelimb exploration), grid-walk (footfaults), open field (locomotor activity), water maze and Y maze (spatial memory), and novel object recognition (NOR, recognition memory) tests were performed as described in the online-only Data Supplement.
Histology and Immunofluorescence
At designated time points, mice were cardioperfused with saline solution and the brains harvested and snap-frozen at −45°C isopentane. Brains were cryosectioned (20-μm-thick slices, 60-μm interval) and stained with luxol fast blue (myelin), hematoxylin and eosin and cresyl violet (neural and infiltrated cells, microbleeds), neurofilament 200 (NF200, axons), CD68 (activated microglia/macrophages), glial fibrillary acidic protein (reactive astrocytes), and Hoechst 33342 (nuclei). Images were acquired using a Nikon Super Coolscan 9000 ED Scanner and a Nikon TE-2000 microscope with epifluorescence illumination, and then analyzed using ImageJ software (NIH).
Infarct Size Measurement
Infarct size was assessed by measuring the volume of demyelinated tissue (luxol fast blue) and was measured using ImageJ. Each section was outlined and the area of injury calculated. The sum of the areas was integrated to obtain injury volume. The volume of corpus callosum/external capsule was also measured and compared among groups to assess possible volumetric differences caused by edema/swelling.
Data are expressed as mean±SD unless otherwise stated. For comparisons between 2 groups, statistical significance was determined using an unpaired Student t test or nonparametric Mann–Whitney test using Prism 5 software (GraphPad Software, San Diego, CA). A P value of <0.05 was considered significant.
Endothelin-1 Microinjections Reduce Blood Flow and Open the Blood–Brain Barrier Into White Matter
There was no mortality after stroke when assessed ≤1 month, and no differences in body temperature were detected between groups during and 2 hours after surgery.
The [14C]-IAP labeling, an index of cerebral blood flow, was significantly reduced at the level of the corpus callosum 4 hours after endothelin-1 microinjection (Figure 1A). Some reduction in flow extended superiorly to grey matter and along the needle tract. Perfusion deficits were not found in vehicle-injected animals (Figure 1B). Extravasation of Evans blue was visible by fluorescent microscopy at the same time point and seemed restricted to subcortical white matter (Figure IA in the online-only Data Supplement).
Endothelin-Induced Ischemia Triggers Neuropathological Alterations in the White Matter
Two days after stroke, myelin degeneration was already visible by a change in magnetization transfer ratio (Figure 2) and by hematoxylin and eosin and luxol fast blue staining (Figure IB in the online-only Data Supplement). Because lesion size was highly variable at this time point,13 we used 1 week after stroke as the main time point for histological assessment. Seven days after endothelin microinjection, focal myelin loss and cellular infiltration were observed at the lesion site by luxol fast blue–cresyl violet and hematoxylin and eosin (Figure 3A and 3D). The majority of animals (>90%) showed a lesion restricted to white matter, sparing both cortical and subcortical grey matter regions. The demyelinated region also showed axonal degeneration as evidenced by the loss of NF200 staining (Figure 3B), indicating tissue damage involving the entire white matter bundle. Because the area of myelin and axonal loss seemed congruent, we performed further volumetric analysis using myelin loss as a surrogate for white matter infarction. Inflammatory infiltrate was identified mainly as CD68+-activated microglia/macrophages by immunofluorescence (Figure 3C). Saline microinjections did not cause any of the above changes (Figure II in the online-only Data Supplement).
White Matter Stroke Is Associated With Long-Term Behavioral Deficit
Behavioral deficit was assessed at 7 and 28 days after stroke in separate cohorts (Figure 4). WT mice receiving endothelin-1 did not show any evident motor dysfunction or contralateral/ipsilateral asymmetry on the cylinder and grid-walk tests when compared with controls (Figure 4A). Neither did locomotor activity differ between groups (Figure 4B) nor did spatial memory using Morris water maze and Y maze (Figure 4C). By contrast, recognition memory was compromised in NOR (Figure 4D; P=0.003), and this persisted for ≥4 weeks after stroke (Figure 4D; P=0.0037). No difference in total exploration time was found between groups.
NOTCH3 Mutation Worsens White Matter Stroke Outcome
There were no obvious developmental differences between WT and N3KO mice at baseline.9 After endothelin-1 microinjection, infarct volume was larger in N3KO mice than in WT animals (Figure 5B; 0.18±0.04 versus 0.10±0.03 mm3, respectively; P=0.0002). No differences in the volume of corpus callosum/external capsule were detected between groups after endothelin-1 injection, mitigating the potential impact of swelling on infarct size (2.05±0.17 versus 2.02±0.19 mm3, respectively; P>0.05). There was minimal or no damage to grey matter adjacent to the injection site. The mutants showed instead greater loss of myelinated fibers into subcortical white matter extending rostro-caudally (Figure 5A) and medio-laterally (Figure IIIA and IIIB in the online-only Data Supplement). When tested at 7 days post stroke, N3KO mice showed a reduced exploratory behavior toward the new object in NOR (Figure 5C; P=0.0125). No differences in baseline NOR (P=0.71; Figure IIIC in the online-only Data Supplement), sensorimotor performances (Figure IIID in the online-only Data Supplement), and total exploration time (NOR, 7.75±0.86 versus 7.69±0.75 s, respectively; P>0.05) were detected among WT and N3KO mice. Analysis of the inflammatory markers, glial fibrillary acidic protein (reactive gliosis), and CD68 (activated microglia/macrophages) did not reveal differences between groups (Figure 6B and 6C; Figure IVA and IVB in the online-only Data Supplement). Microbleeds were not detected when assessed ≤7 days after stroke (Figure 6D; Figure IVC in the online-only Data Supplement).
This study shows that stereotactic microinjections of endothelin-1 cause selective and reproducible ischemic damage to periventricular white matter affecting both myelin and axons. Peptide, but not vehicle, injections prompted a drop in regional blood flow surrounding white matter, causing blood–brain barrier opening, focal cellular infiltration, and histological alteration of corpus callosum, with minimal involvement of adjacent grey matter. Detailed behavioral assessments in these animals did not show deficits in sensorimotor, exploratory, and spatial memory after stroke. However, lesioned mice showed persistent recognition memory impairment. Building on these results, we challenged N3KO mice with white matter stroke to test the hypothesis that NOTCH3 gene deletion affects tissue and functional outcome. Although evaluation longer than 1 week was not possible in the N3KO mice, mutants clearly showed significantly greater infarct size and more pronounced cognitive deficit than WT lesioned animals at 7 days. Moreover, tissue and behavioral deficits persisted for many weeks in the WT, suggesting stability of the lesion and its consequences in this model. Hence, our data show that the endothelin-1 stroke model provides a feasible experimental approach to investigate white matter pathology after small vessel ischemia and demonstrate that NOTCH3 gene deletion increases white matter ischemia susceptibility.
Modeling white matter stroke is challenging. In the past years several animal models have been proposed, but none entirely reproduces the human pathological phenotype.5 One of the main limitations is the small white matter content in rodents14 and the difficulties of restricting the infarct to this small site. Endothelin-1 microinjections can generate small infarcts selectively in white matter and therefore this method offers several advantages specific to brain ischemia. First, endothelin-1 induces a long-term reduction of blood flow in the white matter consistent with ischemic stroke, as already demonstrated in grey matter studies.13,15 Second, endothelin-1 injection causes both myelin and axonal loss resembling human white matter strokes,16 a main advantage over models triggering pure demyelination (eg, lysolecithin injection).17 Third, given the focal lesion obtained with endothelin microinjection, it seems likely that the observed cognitive deficit is mainly from injury to subcortical white matter tracts, ruling out confounding factors (eg, optic nerve damage) affecting global hypoperfusion models.5 Moreover, existing models of small vessel diseases (eg, chronic hypoperfusion)18 are characterized by cumulative vascular impairment leading to tissue and behavioral deficit and reproduce a diffuse, leukoaraiosis-like, white matter damage.7 Differently, the endothelin model more closely resembles the acute pathophysiology of white matter lacunar stroke, although it does not reproduce the actual pathogenesis (eg, occlusion of a small penetrating artery by complications of atherosclerosis and hypertension).3,5,7 This animal model, however, was affected by several technical shortcomings that we needed to improve on, namely variability in lesion size and location, plus obstacles relating to peptide formulation, concentration, and administration. Using the modifications described herein (see Methods in the online-only Data Supplement), we found that peptide microinjections could reliably achieve a focal lesion mainly confined to subcortical white matter and with low variance (infarct size, ≈30% SD). On these bases, we think that the endothelin-1 model can improve our understanding of white matter stroke pathophysiology and facilitate bench-to-bedside translation. However, in its present iteration, it does not mimic the human condition, which is characterized by a complex clinical scenario including multiple infarcts, cognitive deficit, and frontal hypometabolism.4 Nonetheless, the present model may be relevant to mild, early-stage, cognitive impairments associated with lacunar, silent, strokes.3 Indeed, our white matter stroke model triggers episodic-like memory impairment assessed using a novelty-preference task,19 but not a deficit of spatial navigation assigned to hippocampal dysfunction (see below), often considered an indicator of more severe cognitive decline progressing to dementia.20
A major point of this article is that a single white matter lesion can cause a sustained recognition memory deficit, probably because the white matter tracts at the level of the stroke site connect broadly distributed neuronal networks, which coordinate one aspect of cognitive function. A link between white matter lesions, especially at the level of corpus callosum, and impaired cognitive performance has been reported previously. In normal human aging, recognition memory performance depends on the integrity of multiple axonal pathways connecting different brain regions.21 In multiple sclerosis, reduced fractional anisotropy in white matter tracts correlates with cognitive impairment.22 Moreover, periventricular white matter lesions are associated with reduced memory speed and performance, and the severity of the lesion can predict the evolution of cognitive decline.23 In rats, recognition memory is compromised after lesion of the corpus callosum,24 whereas recognition25 and spatial memory26 are impaired after chronic hypoperfusion in mice, along with decreased callosal fiber density. When challenged with chronic hypoperfusion, stroke-prone spontaneously hypertensive rats show diffuse tissue damage (white and grey matter, cortex, and hippocampus) and severe deficit in Morris water maze.27 At variance with these observations, we did not detect impairment in spatial working memory, probably because the anatomic structures (eg, hippocampus)28 or connections relevant for place navigation tasks were spared because of the small size of the lesion. However, we do not rule out the possibility that the endothelin model can generate deficits in spatial memory by modifying size or location of white matter damage (eg, bilateral injections in corpus callosum, infarct location closer to the hippocampus). Nevertheless, here we provide a proof of concept showing that this model can be valuable to better understand the mechanisms underlying the relationship between white mater injury and cognitive deficits.
It has been shown previously that N3KO mice develop greater behavioral deficits and infarct size after middle cerebral artery occlusion compared with WT.9 Here, we demonstrate that the loss of function of NOTCH3 receptor is associated with enhanced white matter ischemia susceptibility after 7 days. The larger infarct size found in N3KO mice did not appear related to grey matter microinfarction or differences in white matter swelling or Willis’ circle anomalies9; reactive gliosis and activated microglia/macrophage infiltration were comparable between groups. Moreover, small hemorrhages and microbleeds, often associated with human NOTCH3 mutations, were not detected in this model, possibly because the human condition is caused by point mutations and frame shifts rather than by gene deletion.8 We speculate instead that the larger infarct can be attributed to greater perfusion deficit in response to the ischemic challenge, similar to previous data from grey matter stroke,9 but this is hard to document with quantitative precision using current autoradiographic techniques because the lesion size is small (≈0.1 mm3) and blood flow is normally low in white matter. Probably the greater behavioral deficit in N3KO mice was because of disruption of a larger number of traversing axons in the larger lesion. Future studies will address to what extent human NOTCH3 mutations (eg, R90C, R169C) affect the ischemic phenotype after white matter stroke and whether these mutants show persistent cognitive impairment.
We are thankful to Dr Elif G. Sozmen in the Laboratory of Dr Thomas S. Carmichael at the University of California Los Angeles for help and advice. We acknowledge the MGH Neuroscience Microscopy and Image Analysis core (P30NS045776).
Sources of Fundings
This work was supported by grants from National Institutes of Health (Dr Moskowitz), Fondation Leducq, Neuroendovascular Research Fund from the Andrew David Heitman Foundation, and The Ellison Foundation (Dr Ayata).
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.005324/-/DC1.
- Received February 27, 2014.
- Revision received February 27, 2014.
- Accepted March 10, 2014.
- © 2014 American Heart Association, Inc.
- Carmichael ST
- Potter GM,
- Marlborough FJ,
- Wardlaw JM
- Bailey EL,
- McCulloch J,
- Sudlow C,
- Wardlaw JM
- Arboleda-Velasquez JF,
- Zhou Z,
- Shin HK,
- Louvi A,
- Kim HH,
- Savitz SI,
- et al
- Zhang K,
- Sejnowski TJ
- Dreier JP,
- Kleeberg J,
- Petzold G,
- Priller J,
- Windmüller O,
- Orzechowski HD,
- et al
- Pavelko KD,
- van Engelen BG,
- Rodriguez M
- Sood R,
- Yang Y,
- Taheri S,
- Candelario-Jalil E,
- Estrada EY,
- Walker EJ,
- et al
- Hort J,
- Laczó J,
- Vyhnálek M,
- Bojar M,
- Bures J,
- Vlcek K
- Shibata M,
- Yamasaki N,
- Miyakawa T,
- Kalaria RN,
- Fujita Y,
- Ohtani R,
- et al
- Jalal FY,
- Yang Y,
- Thompson J,
- Lopez AC,
- Rosenberg GA