Subventricular Zone‐Derived Neural Progenitor Cells Migrate Along a Blood Vessel Scaffold Toward The Post‐stroke Striatum^†

Abstract

The subventricular zone (SVZ) of the adult brain contains neural stem cells that have the capacity to regenerate new neurons after various insults. Brain ischemia causes damage to brain tissue and induces neural regeneration together with angiogenesis. We previously reported that, after ischemic injury in mice, SVZ‐derived neural progenitor cells (NPCs) migrate into the striatum, and these NPCs are frequently associated with blood vessels in the regenerating brain tissue. Here we studied the role of blood vessels during the neural regeneration in more detail. BrdU administration experiments revealed that newly generated NPCs were associated with both newly formed and pre‐existing blood vessels in the ischemic striatum, suggesting that the angiogenic environment is not essential for the neuron‐blood vessel interaction. To observe migrating NPCs and blood vessels simultaneously in damaged brain tissue, we performed live imaging of cultured brain slices after ischemic injury. In this system, we virally labeled SVZ‐derived NPCs in Flk1‐EGFP knock‐in mice in which the blood vessels are labeled with EGFP. Our results provide direct evidence that SVZ‐derived NPCs migrate along blood vessels from the SVZ toward the ischemic region of the striatum. The leading process of the migrating NPCs was closely associated with blood vessels, suggesting that this interaction provides directional guidance to the NPCs. These findings suggest that blood vessels play an important role as a scaffold for NPCs migration toward the damaged brain region. STEM CELLS 2010;28:545–554

INTRODUCTION

Neural stem cells reside in the adult mammalian brain and continuously generate new neurons [1]. A characteristic feature of adult neurogenesis is that new neurons are generated by neural stem cells in restricted regions and then migrate to their final destination, where they function. Therefore, migration is an important step in the neurogenesis of the adult brain [2, 3].

There are two neurogenic regions in the normal adult brain under physiological conditions: the subventricular zone (SVZ) of the lateral ventricle (LV) [4] and the dentate gyrus of the hippocampus [5, 6]. In the SVZ, neural stem cells generate dividing neural progenitor cells (NPCs). In the normal rodent brain, these NPCs migrate tangentially in chains through the rostral migratory stream (RMS) and differentiate only into olfactory bulb (OB) interneurons. However, in the injured brain, young, newly generated neurons appear in the infarct area, where neurogenesis does not occur under normal conditions [7-10]. Stroke induces neurogenesis in the SVZ and the migration of NPCs into the injured striatum [11-14]. Recent studies have implicated several molecular cues in the regulation of NPCs migration in the injured brain, including angiogenic factors, chemokines, and the extracellular matrix (ECM) [15-18]. However, the detailed mechanisms underlying NPCs migration in the injured brain remain unclear.

Vasculature seems to be involved in the stem cell niche of various mammalian tissues [19-21]. In the adult SVZ, neural stem and progenitor cells are observed in the vicinity of blood vessels, which form a vascular niche that maintains the proliferative activity of these cells [22, 23]. In the adult canary brain, brain‐derived neurotrophic factor (BDNF) is produced by newly formed blood vessels in a testosterone‐dependent manner and functions to maintain neurogenesis [24, 25]. In the damaged brain of mammals, including humans, NPCs appear in the vicinity of blood vessels near the injured area [17, 26-28]. We previously reported that chains of generated NPCs are closely associated with blood vessels in the striatum of a mouse model of middle cerebral artery occlusion (MCAO) [13]. In the injured rodent brain, an angiogenic protein and angiogenesis promote the survival of young neurons and accelerate neurogenesis [29-31]. These results strongly suggest a close relationship between the vascular system and neurogenesis in the adult brain. However, the precise roles of blood vessels in neurogenesis or NPCs migration after brain injury remain unclear.

There are two possible functions of the blood vessels during neuronal regeneration after injury. First, newly formed blood vessels may attract NPCs by producing trophic factors that promote the survival and/or differentiation of the NPCs. Second, blood vessels may act as a physical scaffold to guide the migrating NPCs. In this study, we tested these hypotheses using a mouse MCAO model. To examine the role of angiogenesis, we performed long‐term labeling of newly formed blood vessels. We also developed improved techniques for two‐color in vivo cell labeling and live imaging of injured brain slice culture. To overcome technical problems encountered in previous studies, such as cell damage caused by phototoxicity of the emission lamp and the instability of fluorescent dyes for cell labeling, we used an LED light source and genetic cell labeling, respectively. These technological advancements enabled us to observe migrating NPCs and blood vessels in injured brain slice cultures in detail, and for long periods. We show that SVZ‐derived NPCs associate with blood vessels independent of angiogenesis and that they migrate along the blood vessels toward the injured tissue. These results reveal a new role for blood vessels as a scaffold in NPC migration to injured brain tissue.

MATERIALS AND METHODS

Animals

Wild‐type adult ICR and C57Black6 mice (9‐ to 16‐week‐old mice) were purchased from Japan SLC (Shizuoka, Japan). Young heterozygotes of Flk1‐EGFP knock‐in mice (VEGFR‐2 reporter mice) [32] were used for ex vivo culture. On the basis of the national regulations and guidelines, all experimental procedures were reviewed by the Institutional Laboratory Animal Care and Use Committee of Nagoya City University and Keio University and finally approved by the presidents of both universities.

Histological Analysis

Immunohistochemistry was performed as described previously [13]. Detailed conditions are in the supporting information Materials and Methods.

Animal Models of Middle Cerebral Artery Occlusion

Middle cerebral artery occlusion (MCAO) was accomplished using the previously described intraluminal filament technique [13, 33] with modifications. The MCAO protocol is detailed in the supporting information Materials and Methods.

Bromodeoxyuridine Injection Method

To label newly born cells, adult mice were given intraperitoneal (i.p.) injections of bromodeoxyuridine (BrdU; Sigma‐Aldrich, St. Louis, http://www.sigmaaldrich.com) dissolved in phosphate‐buffered saline (50 mg/kg body weight) 2 times a day (every 12 hours) for 0‐18 days after MCAO surgery.

Lentivirus Preparation

Lentivirus was produced as previously described [34]. The lentivirus preparation protocol is detailed in the supporting information Materials and Methods.

Labeling of P1 Neural Stem Cells by Lentivirus Injection

Newborn (P1) mice were anesthetized by hypothermia and placed onto the platform of a stereotaxic injection rig using vinyl tape. A 1‐μl volume of lentivirus was stereotaxically injected into the lateral ventricle [(relative to lambda) anterior, lateral, depth (in mm): 1.8, 1.0, 2.0]. After injection, the mice were placed on a heating bed until the body temperature was elevated.

Live Imaging of NPCs in Ex Vivo Cultured Brain Slices

Live imaging and analysis of NPCs migrating in the organotypic brain slices were performed as reported previously [35, 36], with modifications. Time‐lapse video recordings were obtained using an inverted light microscope (Axio‐Observer; Carl Zeiss, Jena, Germany, http://www.zeiss.com) equipped with the Colibri light‐emitting diode (LED) light system. The live‐imaging protocol is detailed in the supporting information Materials and Methods.

Statistical Analyses

All data were expressed as the mean ± SEM. Differences between means were determined by paired two‐tailed Student's t tests. A p value of <0.01 was considered significant.

RESULTS

Characterization of Angiogenesis and Neurogenesis in the Injured Striatum

We previously reported that newly generated NPCs are associated with some of the blood vessels in the striatum after transient brain ischemia, suggesting that these blood vessels have distinct characteristics [13]. Since ischemia‐induced brain injury induces angiogenesis together with neural regeneration [17, 28-31], the ischemic striatum should contain newly formed blood vessels and also the pre‐existing “old” ones. To label newly formed blood vessels, adult mice were injected with BrdU daily after MCAO and perfused on the day 18, when the generated NPCs were distributed widely in the striatum [13]. To observe the distribution of new blood vessels generated after MCAO, double immunostaining for BrdU and PECAM‐1, an endothelial marker, was performed. BrdU⁺PECAM‐1⁺ endothelial cells were found in the striatum and cortex. In the ischemic striatum of the mice analyzed (n = 18), all of the BrdU⁺PECAM‐1⁺ cells were observed in the area at least 131 μm away from the SVZ. Therefore, the area >131 μm away from the SVZ and the area within 131 μm of the SVZ were defined as angiogenic and non‐angiogenic areas, respectively (Fig. 1A).

Figure 1

Neural progenitor cells (NPCs) are associated with both newly formed and pre‐existing blood vessels in the poststroke striatum. (A): Illustration showing the non‐angiogenic (white) and angiogenic (pink) areas of the coronal brain section 18 days after middle cerebral artery occlusion (MCAO) surgery. (B, C): Triple immunohistochemistry of the striatum with anti‐Dcx (green), anti‐BrdU (red), and anti‐PECAM‐1 (blue) antibodies 18 days after MCAO surgery. (B): NPCs associated with pre‐existing blood vessels in the non‐angiogenic area. (C): NPCs associated with newly formed blood vessels in the angiogenic area. (C′): Cross section of the confocal image of a BrdU⁺ endothelial cell in a newly formed blood vessel, indicated by a dotted line in (C). Scale bar: (B, C) 25 μm. Only 12.4% of the total blood vessels (n = 137) that were associated with Dcx⁺ cells in the injured striatum were labeled with BrdU. Of the 137 blood vessels observed to be associated with NPCs, 59 (43.1%) were in the non‐angiogenic area and 78 were in the angiogenic area. In the angiogenic area, 17 (12.4%) vessels were newly formed and 61 were old blood vessels. Abbreviations: LV, lateral ventricle; SVZ, subventricular zone.

To analyze the NPCs associated with the blood vessels in the ischemic striatum, brain sections were stained for Doublecortin (Dcx), a marker for young neurons, in combination with PECAM‐1 and BrdU. We found that only 12.4% of the total blood vessels (n = 137) that were associated with Dcx⁺ cells in the injured striatum were labeled with BrdU. Of the 137 blood vessels observed to be associated with NPCs, 59 (43.1%) were in the non‐angiogenic area and 78 were in the angiogenic area. In the angiogenic area, 17 (12.4%) vessels were newly formed and 61 were old blood vessels, suggesting that angiogenesis was not required for the neuron‐blood vessel association. Consistent with this idea, in the non‐angiogenic area close to the SVZ, chains of Dcx⁺ cells associated with “old” blood vessels, which did not incorporate BrdU, were observed (Fig. 1B). It is possible that the BrdU labeling protocol used in this study (i.p. injections 2 times a day for 0‐18 days after MCAO) failed to detect some populations of newly formed blood vessels. However, it is very unlikely that the majority of blood vessels associated with NPCs in the “non‐angiogenic area”, which contained no BrdU‐labeled blood vessels, were newly formed blood vessels.

Moreover, to detect angiogenesis in the brain sections, we carefully sought the filopodia of endothelial tip cells, a hallmark of angiogenesis [22, 37]. We frequently found filopodia‐forming tip cells in the ischemic striatum and cortex 3 days after MCAO (data not shown). In contrast, we did not detect them 18 days after MCAO or in the no‐surgery (normal) animal brain (data not shown), indicating that angiogenesis was not active in the brain tissue in the time window during which we analyzed the NPC migration. These data suggest that the association of newly generated NPCs with blood vessels after MCAO is independent of angiogenesis.

To further examine the association between migratory NPCs and non‐newly formed blood vessels, we analyzed the distribution of NPCs after infusing Ara‐C, which induces the regeneration of NPCs without angiogenesis [22]. Indeed, no filopodia‐extending endothelial tip cells were observed in the SVZ 4 days after removal of the Ara‐C infusion pump (data not shown). The antimitotic reagent Ara‐C selectively eliminates dividing cells without affecting nonmitotic cells, including endothelial cells in the normal SVZ, and after the Ara‐C infusion pump is removed, the remaining stem cells regenerate new migrating NPCs rapidly, within 1 week [38, 39]. We found that, in the early stage of NPC regeneration (4 days after removal of the Ara‐C infusion pump), most of the Dcx⁺ NPCs were located in the vicinity of blood vessels (supporting information Fig. 3), indicating that NPCs generated after antimitotic treatment could be associated with pre‐existing blood vessels. These data suggested that, in the initial stages of NPC regeneration, the NPCs' association with blood vessels was independent of angiogenesis.

To characterize the blood vessels associated with NPCs in more detail, we measured their diameter. Their average diameter was 7.7 ± 0.2 μm (n = 137 blood vessels). We also stained the ischemic brain sections for Aquaporin4 (AQP4), a marker of astrocytic endfeet [40]. We frequently found thin astrocytic processes expressing AQP4 between the Doublecortin⁺ (Dcx) NPCs and blood vessels (supporting information Fig. 1). These data suggested that astrocytic endfeet are involved in the interaction between the NPCs and blood vessels, consistent with our previous observations using electron microscopy [13].

Live Imaging of Virally Labeled NPCs Migrating from the SVZ Toward the Injured Striatum of Flk1‐EGFP Knock‐in Mice

A number of recent studies have suggested that blood vessels play important roles in neurogenesis in the adult brain [20, 22, 23, 25, 41, 42]. An association of newly generated NPCs with blood vessels has been reported in various animal models of brain injury and even in human patients after stroke [17, 26-28]. Previous studies using fixed brain sections strongly suggested that the NPCs use blood vessels as a physical scaffold for their migration in the injured brain [13, 17], and one recent study demonstrated that NPCs migrate along blood vessels in the striatum after stroke [26]. However, the functional relationship between the blood vessels and newly generated NPCs has not been demonstrated clearly. To observe the movement of the regenerated neurons along blood vessels in live brain tissue in more detail, we performed live imaging of the migrating NPCs and blood vessels in injured brain slices (Fig. 2).

Figure 2

Procedure for time‐lapse imaging of injured brain slices. (A): Experimental design for imaging neural progenitor cells (NPCs) migrating in injured brain slices. A lentiviral vector encoding DsRed‐Express was injected into the lateral ventricle of P1 Flk1‐EGFP knock‐in mice to label periventricular neural stem cells and their progenies. Thirteen days after viral injection, the P14 mice were subjected to MCAO surgery. TTC staining demonstrated a hypoxic‐ischemic region in the striatum and cortex (indicated by small blue arrows) of the brain after MCAO surgery. Then, 10‐15 days after MCAO surgery, coronal brain slices were cut and cultured. Time‐lapse recordings of the DsRed‐Express‐expressing cells and EGFP‐expressing blood vessels in cultured brain slices were obtained using LED light microscopy. (B, C, E): Coronal sections of a normal P25 mouse brain (B) and brains 13 days after MCAO. (C, E): Stained with an anti‐Dcx antibody. The region within the box in (C) is enlarged in (D). (D, D′, D″): Higher magnification of the boxed region in (C). Chains of Dcx⁺ (green) cells were associated with PECAM‐1⁺ (red) blood vessels in the striatum. (E): Dcx⁺ NPCs with a leading process pointing to the SVZ is indicated by a green arrow. (F–H): Coronal section of a brain 8 days after the injection of a lentiviral vector encoding DsRed‐Express into the lateral ventricle. In the SVZ, a subpopulation of DsRed‐Express⁺ cells (red) expressed Dcx (green). (I–K): Coronal sections of the brain of a P25 Flk1‐EGFP knock‐in mouse stained with an anti‐PECAM‐1 antibody (red). GFP signals (green) were colocalized with PECAM‐1⁺ blood vessels. Scale bars: (B, C) 100 μm; (D–D″) 50 μm; (E) 50 μm, (F–H) 50 μm, (I‐K) 100 μm. Abbreviations: GFP, green fluorescent protein; LED, light‐emitting diode; MCAO, middle cerebral artery occlusion; LV, lateral ventricle; SVZ, subventricular zone.

In general, it is difficult to perform long‐term recording of NPCs migration in cultured adult brain slices, because aging decreases the cell viability in culture. Therefore, most of the previous live‐imaging studies of neuronal migration used embryonic or neonatal brain [35, 43, 44]. Thus, we first designed a stroke model using young mice. For this purpose, we modified our adult MCAO protocol [13, 33] to create a stroke model in younger mice (P14). We found that transient occlusion of the middle cerebral artery with a filament for 40 minutes caused ischemic injury in the striatum and cortex but not in the SVZ, as revealed by TTC staining [45] 3 hours after surgery (Fig. 2A, small blue arrows). We next studied the distribution of NPCs after MCAO in this new stroke model. Ten days after MCAO, the mice were killed and analyzed. In the normal control brain of the same age (P24), Dcx⁺ cells were restricted to the SVZ (Fig. 2B). In the brain of MCAO‐treated mice, Dcx⁺ cells were present in the striatum and the SVZ, and in the injured striatum, some of them formed a chain‐like structure (Fig. 2C, 2E). Double staining for Dcx and PECAM‐1 revealed that some of the Dcx⁺ cells were associated with blood vessels in the injured striatum (Fig. 2D). These observations are similar to previously reported results using an adult MCAO model [13].

To visualize blood vessels and migrating NPCs simultaneously, we needed to label them with fluorescent proteins of different colors in our young MCAO model. We injected lentiviral vectors encoding DsRed‐Express into the lateral ventricles of P1 mice to label the radial glial cells located in the neonatal SVZ, the origin of adult neural stem cells [46]. Abundant DsRed‐Express⁺ cells were detected in the SVZ of P8 mice 7 days later (Fig. 2F). A subpopulation of these DsRed‐Express⁺ cells was positive for Dcx (Fig. 2F–2H), indicating that this method could label the NPCs generated in the SVZ of young mice. To label all of the blood vessels stably, we used Flk1‐EGFP knock‐in mice, which express EGFP in endothelial cells [32]. The immunostaining of brain sections for PECAM‐1 confirmed that all of the PECAM‐1‐expressing endothelial cells were labeled with EGFP. The EGFP signals were completely colocalized with the PECAM‐1, and no ectopic signals were detected in the young mouse brain (Fig. 2I–2K), indicating that the Flk1‐EGFP knock‐in mice were suitable for visualizing live blood vessels in the slice culture of the young MCAO model.

We injected the lentiviral vector encoding DsRed‐Express, prepared slice cultures from control mice without MCAO, and performed time‐lapse recording using an LED light, which is less toxic to cells than a mercury lamp [47-49]. In the control experiments using normal brain slices, during a 20‐hour recording period, DsRed‐Express⁺ cells were observed to move within the SVZ, and none migrated into the striatum from the SVZ (four brain slices) (supporting information Movie 1).

We next performed the live imaging of brain slices after MCAO. Since long exposure to two excitation lights caused severe damage to cells in slice culture (data not shown), we first used a single‐color fluorescent light for the long‐term recording of migrating NPCs (∼30 hours). The results showed that DsRed‐Express⁺ cells migrated out from the SVZ and moved toward the injured striatum (Fig. 3A, supporting information Movie 2). Migrating NPCs elongated their process toward the direction of migration (Fig. 3A). To investigate the direction and speed of the cell migration, we traced the trajectory of the moving cells (Fig. 3B). In the long‐term imaging analyses, labeled cells were observed to migrate out from the SVZ and to move continuously toward the ischemic striatum for approximately 691.2 μm (Fig. 3B, indicated in red). We also observed that 55.6% of the cells made a U‐turn (five out of a total of nine migrating cells); after first migrating toward the ischemic striatum, they turned back to the SVZ (Fig. 3B, indicated in green). The average migration speed was 33.1 ± 1.6 μm/h (n = 9 cells in 5 slices).

Figure 3

Live imaging of neural progenitor cells (NPCs) migrating over a long distance in slices of injured brain. (A): Time‐lapse series of DsRed‐Express⁺ cells migrating in slices of injured brain. The insets are high‐power magnification views of the small boxed region in the same image. The movement of two DsRed‐Express⁺ cells (indicated by green and red arrows) was monitored for 14 hours. One cell (green arrow) migrated out from the SVZ toward the ischemic striatum and then turned back toward the SVZ. The other cell (red arrow) continuously migrated toward the ischemic striatum throughout the entire recording period. (B): Schematic drawing of the migration pathway of two DsRed‐Express⁺ cells. The total migration distances were 692.1 μm (green) and 691.2 μm (red). Scale bar: (A, B) 100 μm. Abbreviations: LV, lateral ventricle; SVZ, subventricular zone.

We next examined whether NPCs would exhibit this U‐turn in vivo and also in vitro. In fixed brain sections, we examined the leading process of Dcx⁺ NPCs, which is oriented in the direction of migration. As expected, the leading process of individual NPCs pointed in various directions, including toward the SVZ (indicated by an arrow in Fig. 2E), suggesting that NPCs frequently turn back to the SVZ from the striatum in vivo.

Next, we investigated whether the migration of NPCs in the injured brain slices was affected by CXCL12 (SDF‐1)/CXCR4 signaling, which has been reported to regulate NPC migration in MCAO models [15, 18]. AMD3100 is a specific and potent small‐molecule inhibitor of CXCR4 that decreases NPC migration toward the injured striatum in vivo [18, 50-52]. We added AMD3100 to the culture medium of the injured brain slice and recorded the NPC migration. AMD3100 significantly decreased the migration speed of the NPCs (control, 32.7 ± 1.7 μm/h, n = 7; AMD3100, 6.6 ± 0.5 μm/h, n = 5; p = .000000282). All of the NPCs observed were restricted to the region within 103 μm of the SVZ (supporting information Fig. 2, supporting information Movie 4). These data indicate that CXCR4 signaling is important for NPC migration in the injured brain. This experiment also shows that our ex vivo live‐imaging system is useful for elucidating the molecular mechanisms of NPC migration in the injured brain.

Finally, to examine the relationship between blood vessels and the migrating cells in the injured striatum, we performed time‐lapse imaging of DsRed‐Express⁺ cells and EGFP⁺ blood vessels using two‐color excitation LED light. We looked for DsRed‐Express⁺ cells with a process extending toward the ischemic region in the vicinity of blood vessels and performed time‐lapse imaging for 5 hours. The cell body and leading process of DsRed‐Express⁺ cells (n = 7) were observed to move along blood vessels labeled with EGFP (supporting information Movie 3, Fig. 4). The changes in cell morphology recorded in this movie suggest that the DsRed‐Express⁺ cell retracted its leading process and reversed its direction when the tip of the leading process reached the branch point of a blood vessel (n = 1). This observation suggests that the vascular network may influence the NPC migration during neuronal regeneration after brain injury.

Figure 4

Neural progenitor cells (NPCs) migrate along blood vessels in the ischemic striatum. Time‐lapse images of NPCs migrating in the ischemic striatum. (B, C): Higher magnification of the boxed region in (A). (B): A DsRed‐Express⁺ cell (white) migrated toward the ischemic striatum (0‐90 minutes), pointing its leading process (arrow) in the direction of migration. Subsequently, the cell retracted the leading process (120‐150 minutes), extended another process in the opposite direction (180‐210 minutes), and turned back toward the SVZ. (C): Two‐color time‐lapse analysis of the migrating NPCs (green) and the blood vessels (red). Note that the leading process of the NPCs was associated with a blood vessel (indicated by arrows) during its migration. (D): Illustration of the NPCs migration shown in (C). Scale bars: (A) 100 μm; (B–D) 50 μm. Abbreviations: LV, lateral ventricle; SVZ, subventricular zone.

DISCUSSION

Improved Method for Live‐Imaging Studies of Cell Migration in the Injured Brain

In the past decade, the time‐lapse recording of cells in slice culture has successfully revealed mechanisms of neuronal migration in the fetal or neonatal brain [35, 43, 44]. However, most of the previous studies on NPCs migration in the injured adult brain were performed using histological analyses of fixed brain tissue [13, 14, 16, 17, 28], which cannot reveal the detailed behavior of NPCs. Although recent short‐term studies with brain slices have revealed some characteristics of live, newly generated NPCs in the injured brain, an extension of the recording time is critical for tracing NPCs migrating a long distance from the SVZ. A recent paper reported the live imaging of newly generated NPCs labeled with DiI in slice culture of the injured adult brain [53]. However, the injection of DiI solution into lateral ventricles results in limited labeling of the NPCs and strong labeling of cells lining the lateral ventricles [54], which prevents a detailed observation of the migrating NPCs. Another paper recently reported the time‐lapse recording of migrating NPCs in ischemic adult brain slices from Dcx‐EGFP transgenic mice in which all of the Dcx‐expressing NPCs are visualized [26]. However, it is difficult to observe the morphology and behavior of migrating NPCs in Dcx‐EGFP transgenic mice at the single‐cell level because most of the EGFP‐expressing NPCs form aggregates in the striatum and in the SVZ. In addition, the EGFP‐expressing cells in the striatum may include some other cell populations in addition to the SVZ‐derived NPCs.

To overcome these technical limitations and to record specifically the behavior of SVZ‐derived NPCs migrating for a long distance at high resolution, we observed virally labeled red‐fluorescent NPCs migrating in cultured brain slices prepared from an MCAO model of young knock‐in mice carrying a blood vessel‐specific GFP reporter. Because radial glial cells, the origin of adult neural stem cells, line the ventricular walls of neonatal mice [46, 55], the injection of a DsRed‐Express‐encoding lentivirus into the lateral ventricles of P1 mice enabled the efficient labeling of cells derived from the SVZ.

Fluorescent dye injection, which has been used to visualize live blood vessels in previous studies [26, 56, 57], has some technical limitations, such as incomplete and unstable labeling. To improve the efficiency of labeling and the imaging resolution for long‐term slice culture, we used Flk1‐EGFP (VEGFR‐2) knock‐in mice, which express EGFP in all the endothelial cells in blood vessels, from the embryonic to adult stage.

Finally, to minimize the damage to cells during recording, we used LED as the light source. LED light has less phototoxicity for live cells compared with that of other conventional light sources, such as a mercury or xenon lamp, because it emits less UV and infrared radiation [47-49]. These modified approaches enabled us to increase the recording time of NPCs migrating in the slices of injured brain.

Nevertheless, there are at least three differences between the conditions of in vivo tissues and ex vivo cultures. First, the nutrient and oxygen supply can become insufficient in slice culture, causing cell death during long incubation periods spanning ∼20 hours. Second, there is no blood flow in cultured slices, so the effects of blood flow and blood‐derived factors on NPCs migration cannot be studied in this system. Third, the movement of cells is spatially limited in slice culture. Our brain slices were about 200‐μm thick, so the NPCs migration perpendicular to the plane of the slice surface was limited. Any of these differences could have affected the behavior of the newly generated NPCs. Furthermore, the NPCs migration in the young brains used in this study might be different from that in the adult brain. However, we carefully confirmed that the movement of labeled cells was restricted within the SVZ in the normal brain slice culture and that cell migration toward the striatum was observed only in the tissue from injured brains. Furthermore, the small‐molecule inhibitor‐mediated suppression of CXCL12/CXCR4 signaling, which was previously reported to be involved in NPC migration in the injured brain by the analysis of fixed tissue [18], decreased the speed of NPC migration in our live‐imaging system. Therefore, we conclude that the behavior of NPCs migrating toward the injured area presented in this paper was induced by the ischemic stimuli, and not an artifact of the ex vivo culture.

The Migration Pattern of Newly Generated NPCs in the Injured Brain

Using our novel experimental setting, we studied the speed and direction of the newly generated NPCs. Since previous studies demonstrated that the regeneration efficiency of neurons in the injured brain is extremely low (about 0.2%) [14], we expected the migration speed of newly generated NPCs to injured tissue to be slower than that of NPCs in the RMS under normal conditions, because of limited guidance cues and/or physical scaffolds. However, the migration speed of NPCs observed in the striatum was 33.1 ± 1.6 μm/h, which is comparable to that of NPCs in the RMS (98 μm/h [35]) and OB (50 μm/h; data not shown) in the normal adult brain. We previously reported that newly generated striatal NPCs exist mainly 40‐400 μm away from the SVZ [13]. Therefore, these NPCs seem to reach their final site of differentiation in the striatum within 24 hours. A previous time‐lapse imaging study demonstrated that NPCs expressing a DCX‐EGFP reporter gene moved ∼150 μm in the striatum of the ischemic brain [26]. The present study provides direct evidence that NPCs generated in the SVZ migrate for a long distance spanning over 691.2 μm (Fig. 3, supporting information Movie 2).

Interestingly, we observed that NPCs migrating toward the striatum frequently turned back toward the SVZ. Similar exploratory behavior of generated NPCs was previously reported [26]. Consistent with these observations, we also observed Dcx⁺ NPCs in the striatum, pointing their leading process toward the SVZ in fixed brain tissue, suggesting that similar U‐turn behavior takes place in vivo. In our previous study, almost all the newly generated NPCs remained within 400 μm from the SVZ [13]. The tendency of these NPCs to turn back to the SVZ might be one of the reasons why they migrate only a limited distance from the SVZ and do not reach the ischemic core region, in spite of their high speed.

Function of Blood Vessels in NPCs Migration in the Ischemic Striatum

Previous studies have demonstrated that migrating NPCs are associated with blood vessels in the ischemic striatum [13, 17, 26, 28]. Using improved cell‐labeling and live‐imaging methods, we clearly showed here that the SVZ‐derived migrating NPCs extend their leading process along blood vessels that are oriented in the direction of the ischemic region and then move their cell bodies along the blood vessels. Interestingly, NPCs that were migrating toward the ischemic region reversed their direction when the tip of the process touched the branch of a blood vessel (Fig. 4C). In the injured brain, most NPCs migrated only a short distance from the SVZ. Our observation that NPCs turn back when they hit the branch point of a blood vessel might be one of the reasons why only a few NPCs eventually reach the ischemic site.

The mechanisms underlying the migration of newly generated NPCs along blood vessels in the injured brain are unknown. One possibility is that some attractive molecules derived from the blood contribute to the NPC‐blood vessel interactions during NPCs migration. Since newly formed blood vessels in the injured brain are defective in the blood‐brain barrier, the NPCs in the vicinity of newly formed blood vessels might easily be exposed to blood‐derived molecules. However, we found that newly generated NPCs were associated with both “old” and newly formed blood vessels. Consistent with this observation, recent studies have demonstrated that NPCs migrate along blood vessels in the RMS and OB of the normal adult brain [56-58], where angiogenesis rarely occurs. Recent reports also showed that the formation of the blood‐brain barrier in the blood vessels of the SVZ is incomplete, suggesting that NPCs might receive vasculature‐derived molecules in the normal SVZ [22, 23]. However, since this type of modified blood‐brain barrier has not been found in other areas, including the striatum, blood‐derived factors might not be required for the association of NPCs and blood vessels in the injured brain observed in the present study.

Recent studies have indicated that the migration of regenerated NPCs and angiogenesis are coupled [17, 29]. After ischemic injury, an increased expression of the angiogenic factors, such as Angiopoietin1, occurs in the injured area, and these factors induce angiogenesis via receptor molecules expressed in endothelial cells, such as Tie2. These receptors are also expressed in NPCs and play a role in regulating their migration. The angiogenesis‐mediated regulation of NPC migration is well known to play important roles in the brain, as demonstrated in various stroke models [17, 29]. However, it is unlikely these factors play a major role in the blood vessel‐guided migration of NPCs observed in this study, for the following reasons. MCAO induces angiogenesis within 10 days of the injury in the rodent brain [31, 59], but we analyzed the relationship between NPCs and blood vessels 18 days after the injury, when the angiogenic environment should be diminished. Furthermore, a similar association of NPCs with blood vessels is observed in the nonischemic brain after Ara‐C treatment, which eliminates dividing cells without affecting the pre‐existing blood vessels [22, 38, 39], suggesting that the newly generated NPCs are associated with blood vessels independent of angiogenesis (supporting information Fig. 3). Consistent with this idea, newly generated NPCs are continuously supplied to the injured area a year after the injury in a rat MCAO model [28]. We demonstrated that AMD3100, an inhibitor of CXCL12/CXCR4 signaling, suppressed the NPC migration (supporting information Fig. 2, supporting information Movie 4). In addition to angiogenesis, this signaling has many functions, such as in the mobilization of hematopoietic stem cells, the infectivity of HIV, and neuronal migration [60-63]. Therefore, the AMD3100‐induced inhibition of NPC migration is not necessarily attributable to impaired angiogenesis.

ECM is abundant around the blood vessels in the brain and has been suggested to be involved in cell survival and migration [64-66]. Laminin is a major component of the ECM around blood vessels in the SVZ, and NPCs express the laminin receptor α6β1 integrin. The blockage of α6β1 integrin inhibits the association of NPCs with blood vessels [23]. A recent study showed that BDNF derived from blood vessels promotes NPCs migration via the p75NTR receptor in the RMS under physiological conditions [57]. Thus, these molecules may be involved in the control of NPCs migration along blood vessels toward the injured tissue.

Why do the NPCs use blood vessels as a scaffold for their migration toward the injured area? In the SVZ and RMS of the normal adult brain, NPCs show chain migration [67] in which they move in association with each other inside glial tunnels, which enables their efficient migration in mature brain tissue. In the developing cerebral cortex, new neurons use the fiber of radial glial cells as a scaffold for their migration [55, 68-70]. Since the adult striatum does not contain radial fibers or glial tubes, the newly generated NPCs need some other physical scaffold. Moreover, blood vessels are uniformly distributed throughout the striatum and are suitable as a scaffold for migrating NPCs in the injured brain. Blood vessels have been recently reported to function as a physical scaffold in the RMS and OB of the normal brain [56, 57], indicating that this mechanism is common to both regenerative and physiological neurogenesis (Fig. 5).

Figure 5

Roles of blood vessels in neural progenitor cells (NPCs) migration in the adult brain. In the normal brain, NPCs (green) migrate from the SVZ to the OB via the RMS. After a stroke, the SVZ‐derived NPCs migrate from the SVZ toward the injured area (purple). Blood vessels act as a scaffold for the migration of SVZ NPCs under both normal and injured conditions. Abbreviations: LV, lateral ventricle; OB, olfactory bulb; RMS, rostral migratory stream; SVZ, subventricular zone.

CONCLUSION

In summary, this work has demonstrated that, after MCAO in mice, SVZ‐derived NPCs show complex migratory behavior along a blood vessel scaffold, and the association of NPCs with blood vessels is independent of angiogenesis in the injured brain tissue. In addition, the live‐imaging method for NPCs migration in the injured brain developed in this work will be useful for studying the detailed mechanisms underlying the guidance of NPCs.

Acknowledgements

We are grateful to H. Miyoshi for the lentivirus constructs, H. Naka, T. Yamashita, M. Ninomiya, and H. Ichijo for technical advice and members of the Sawamoto laboratory for discussions and encouragement. This work was supported by the Ministry of Education, Culture, Sports, Science and Technology, the Japan Society for the Promotion of Science, the Ministry of Health, Labor and Welfare, the Toray Science Foundation, the Kanae Foundation for the Promotion of Medical Science, the Takeda Science Foundation, Keio University Medical Science Fund, Mitsui Life Social Welfare Foundation, Global COE Programs at Keio University, and Keio University Grant‐in‐Aid for Encouragement of Young Medical Scientists. T.K. is a Japan Society for the Promotion of Science Research fellow.

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors indicate no potential conflicts of interest.