Author contributions: N.H.: design, collection, and assembly of data, data analysis and interpretation, manuscript writing; M.P.P.: design, collection, and assembly of data, data analysis and interpretation, manuscript writing; K.N.: collection and assembly of data, data analysis and interpretation; G.S.: collection of data; H.K.-G.: collection of data; A.E.: collection of data; M.S.: collection of data; S.A.H.: statistical analysis advice; K.J.: collection of data; S.H.: provision of study material, manuscript writing; T.D.P.: data analysis and interpretation; T.M.B.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript; G.K.S.: financial support, design, data interpretation, manuscript writing, final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS January 14, 2011.
Cell transplantation offers a novel therapeutic strategy for stroke; however, how transplanted cells function in vivo is poorly understood. We show for the first time that after subacute transplantation into the ischemic brain of human central nervous system stem cells grown as neurospheres (hCNS-SCns), the stem cell-secreted factor, human vascular endothelial growth factor (hVEGF), is necessary for cell-induced functional recovery. We correlate this functional recovery to hVEGF-induced effects on the host brain including multiple facets of vascular repair and its unexpected suppression of the inflammatory response. We found that transplanted hCNS-SCns affected multiple parameters in the brain with different kinetics: early improvement in blood-brain barrier integrity and suppression of inflammation was followed by a delayed spatiotemporal regulated increase in neovascularization. These events coincided with a bimodal pattern of functional recovery, with, an early recovery independent of neovascularization, and a delayed hVEGF-dependent recovery coincident with neovascularization. Therefore, cell transplantation therapy offers an exciting multimodal strategy for brain repair in stroke and potentially other disorders with a vascular or inflammatory component. STEM CELLS 2011;29:274–285
Stroke is a leading cause of long-term disability with very few therapeutic options. Because of its complex pathology including damage to neurons, glia, and endothelial cells in the brain, conventional therapeutic strategies target the first few critical hours after stroke onset to minimize stroke-induced damage. Cell transplantation presents a novel therapeutic approach with the potential to repair the damaged brain and therefore extend the therapeutic time window of intervention, thus benefiting significantly more stroke patients. A diverse array of transplanted cell types, including brain-, bone marrow-, and blood-derived progenitors are reported to enhance functional recovery after stroke [1–6], and several cell transplantation clinical trials for stroke are currently underway (clinicaltrials.gov Identifier: NCT00473057; NCT00859014; NCT00535197; NCT00950521. 2009). The cells used in this study, that is, human central nervous system stem cells grown as neurospheres (hCNS-SCns), are a potentially exciting candidate for stroke therapy as they are currently in clinical trials for several other central nervous system disorders (http://www.stemcellsinc.com).
Despite multiple reports indicating that stem cell transplantation is beneficial after stroke, the mechanisms of stem cell-induced recovery are poorly understood and may differ depending on the cell type studied. Secretion of trophic factors by transplanted cells is speculated to be a major contributor to their beneficial effects, but it is not known which factors are necessary to elicit recovery. Several studies have overexpressed factors in transplanted stem cells and found recovery was further enhanced [7, 8]; however, such experiments do not elucidate whether these factors are sufficient to stimulate recovery or whether they can only amplify recovery in an already primed system. Therefore, identification of crucial stem cell-secreted factors remains to be determined. Furthermore, it is not understood what changes occur in the brain in response to the grafted stem cells, the role of stem cell-secreted factors in these changes, or how these changes relate to stem cell-induced recovery; understanding such a cause and effect relationship will be imperative to understanding the mechanism of action of transplanted cells.
In this study, we begin to address these questions by selectively neutralizing vascular endothelial growth factor (VEGF) secreted by transplanted hCNS-SCns and investigating how this affects functional recovery and various stem cell-induced changes in the poststoke brain. We chose to study VEGF because it is a key proangiogenic factor and increased vascularization and perfusion in the peri-infarct region within a few days after stroke is associated with neurological recovery in patients [9, 10]. Moreover, acute transplantation of bone marrow- or blood-derived cells after stroke enhances blood vessel formation and, in some studies, functional recovery in rodents [11–13]. It is therefore postulated that stem cell-induced vascularization after stroke is important for cell-induced recovery [14, 15]. Additionally, the tight network of communication between the vasculature and the neurovascular unit, which is composed of neurons, astrocytes, and microglia , implies that effects on the vasculature have the potential to significantly influence brain function . Inflammation, another major determinant of stroke pathology, can also affect vascularization and blood-brain barrier (BBB) integrity through release of proangiogenic factors and reactive oxygen species [17, 18], and there is growing evidence that interactions between the neurovascular unit and inflammation are also critical to stroke recovery . Moreover, stem cell transplantation is reported to decrease inflammation in rodent models of stroke and multiple sclerosis [20–24], but it is not understood how.
In summary, this study investigates for the first time the in vivo role of a stem cell-secreted factor in mediating functional recovery in the stroke brain. We neutralize VEGF secreted by transplanted hCNS-SCns, determine how this alters stem cell-induced functional recovery, and relate this to changes in stem cell-mediated effects on vascular regeneration including neovascularization, restoration of BBB integrity, and neuroinflammation, which are all postulated to significantly influence poststroke recovery.
MATERIALS AND METHODS
Distal Middle Cerebral Artery Occlusion and Cell Transplantation
Animal procedures were approved by Stanford University's Administrative Panel on Laboratory Animal Care. T cell-deficient adult male Nude rats  (Cr:NIH-RNU 230 ± 30 g) were subjected to permanent distal middle cerebral artery occlusion (dMCAo) with 0.5 hours bilateral CCA occlusion as described  under isoflurane anesthesia. Ampicillin was administered in the drinking water (1 mg/ml) starting 3 days before stroke surgery to 7 days post-transplantation surgery.
Fetal-derived hCNS-SCns  were dissociated for transplantation . Three 1.0 μl cell deposits (1 × 105 cells per microliter) or vehicle were injected into the ipsilesional cortex at 7 days post-dMCAo as described : (a) anterior–posterior (A–P), +1.6; medial–lateral (M–L), −2.4; dorsal–ventral (D–V), −2.4; (b) A–P, +0.7; M–L, −2.4; D–V, −2.4; (c) A–P, −0.3; M–L, −2.4; D–V, −2.4. For “dead cell” transplants, the cells went through four cycles of freeze thaw using dry ice and a warm water bath, prior to transplantation.
Tissue Processing, Measurement of Cortical Atrophy, and Cell Survival
Rats were perfused and 40-μm coronal sections prepared . Cortical atrophy at 4 weeks post-transplantation was measured (blinded) in cresyl violet-stained sections by a semistereological approach. Animals were chosen at random from each group. Five serial sections, 0.5 mm apart starting at A-P + 1.7, were taken per brain, as this encompassed the lesion. The remaining ipsilesional cortex was measured and divided by the area of the contralateral cortex. hCNS-SCns survival in animals chosen randomly at 2 weeks post-transplantation was determined (blinded) by stereological counting as previously described .
Immunohistochemistry and Inflammation Analysis
Primary antibodies were incubated overnight, 4°C : SC121 (1:500; human cytoplasmic marker, StemCells, Inc., Palo Alto, California, www.stemcellsinc.com/), anti-β-dystroglycan (βDG; 1:100, Novocastra Laboratories Ltd, Newcastle upon Tyne, United Kingdom, http://www.ebiotrade.com/buyf/Novocastra/INDEX.HTM), rabbit anti-aquaporin 4 (1:500, Sigma-Aldrich Corp., St. Louis, MO, http://www.sigmaaldrich.com/united-states.html/), anti-human VEGFa (1:100, Abcam, Cambridge, United Kingdom, http://www.abcam.com/), anti-ionized calcium binding adaptor molecule 1 (Iba1) (1:1,000, Waco Chemicals USA, Inc., Richmond, VA, http://www.wakousa.com/), Cy3 F(ab′)2 Frag donkey anti-human IgG (1:500, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, http://www.jacksonimmuno.com/), anti- β-III tubulin (1:1,000, Millipore, Billerica, MA, http://www.millipore.com/), anti-nestin (1:250, Abcam), and anti-human nuclei (1:100, Millipore). Secondary antibodies were incubated 2 hours at room temperature (1:1,000, Alexa Fluor 568 or 488, Jackson ImmunoResearch), with DAPI (1:1,000, Calbiochem, EMD4Biosciences Inc., Gibbstown, NJ, http://www.emdchemicals.com/), followed by confocal microscopy analysis. Cell lysates were measured for human VEGF (hVEGF) by enzyme-linked immunosorbent assay (ELISA) using the Quantikine human VEGF kit (R&D Systems, Minneapolis, MN, http://www.rndsystems.com/).
Stereological quantification of Iba-1 stained microglia/monocytes was done as described . Four sections per animal (each 40 μm apart spanning A–P 1.0–0.5 mm encompassing the majority of the lesion) were blindly analyzed using Stereo Investigator (MBF Bioscience, Williston, VT, http://www.mbfbioscience.com/) and average density determined. Two regions of interest (ROIs) were counted per slice, the peri-infarct region (defined in Fig. 6A), and the remainder of the ipsilesional cortex. Two sections per animal were counted for Avastin-treated and IgG-treated animals; this gave similar trends to counting four sections.
Blood Vessel Density Analysis
Vessels were labeled by jugular vein injection of fluorescein 5′-isothiocyanate (FITC)-lectin (1.6 mg/kg, Vector) 30 minutes presacrifice. Blood vessel density (BVD) was measured (blinded) using Image J to determine pixel number per image. The ratio of ipsilateral to contralateral cortical BVD was calculated. BVD was analyzed in three ROIs: the cortical ischemic penumbra; near the graft; between the two, and in equivalent ROIs in the contralateral cortex. In each of the ROIs, three sub-ROIs were imaged to sample vessels from different cortical layers. Three sections were analyzed per animal in the same A–P region defined above. Thus, for each ROI, nine images were analyzed per hemisphere and the average of these BVD ratios taken. Vessel width was also measured as it could affect pixel number. There was no significant difference in average vessel width in the absence of antibody treatment. Animals receiving intraperitoneal injections had nonspecific changes in vessel width due to the resultant inflammatory response; therefore, their BVD was normalized for vessel width. To confirm increased vascularization, total vessel length was also measured per image.
Measurement of BBB Permeability
Vessel leakage was evaluated by jugular vein injection of sulfo-NHS-biotin (0.07 mg/g, Pierce) 30 minutes presacrifice. Four coronal sections were stained with streptavidin-conjugated Alexa 555 or Cy-5 (1:500, Jackson Immunoresearch Laboratories, Inc., West Grove, PA, http://www.jacksonimmuno.com/) and leakage area measured on confocal images using a Zeiss LSM Image browser.
Sequestration of Human VEGF with Avastin
A total of 4 mg/kg Avastin (Bevacizumab, Genentech, South San Francisco, California, http://www.gene.com/gene/contact/locations.html) in saline was injected into the peritoneal cavity every other day  starting transplantation day. Human IgG (Sigma) was used for control injections.
Microvessel Isolation and Western Blot
Ipsilesional cortical tissue was minced (after removing meninges and large surface vessels), mixed with 30% dextran (Sigma) in Earle's balanced salt solution (EBSS; GIBCO, Grand Island, New York, http://www.invitrogen.com/site/us/en/home.html/), centrifuged (3,220g, 10 minutes, 4°C), the pellet washed in EBSS, repelleted, resuspended in 1 ml EBSS, and passed through a 70-μm then 40-μm cell strainer (BD Falcon, http://www.bdbiosciences.com/home.jsp/). Microvessels were collected from the 40-μm strainer membrane, washed in EBSS, repelleted at 110g, resuspended, and sonicated in 1× lysis buffer (Cell Signaling Technology, Danvers, MA, http://www.cellsignal.com/) containing phosphatase (Sigma) and protease inhibitors (Roche, Basel, Switzerland, http://www.roche.com/index.htm/). A total of 1–5 μg of protein (equal loading per gel, per experiment) underwent SDS polyacrylamide gel electrophoresis (PAGE) using 3%–8% NuPAGE Tris-acetate gels (Invitrogen, Carlsbad, California, http://www.invitrogen.com/site/us/en/home.html/) and the gels blotted to polyvinylidinene difluoride membranes (Invitrogen). The membranes were incubated with primary antibody, 4°C overnight: rabbit anti-phospho VEGFR2 (1:500, Millipore), rabbit anti-Tie 2 (1: 2,000, Calbiochem), mouse anti-Occludin (1:500, BD Biosciences), rabbit anti Zo-1 (1:250, Invitrogen), mouse anti-Claudin-5 (1:3,000, Lifespan BioSciences, Seattle, WA, http://www.lsbio.com/), mouse anti-β actin (1:10,000, Sigma). Secondary anti-mouse (1:2,000) or anti-rabbit antibodies (1:3,000, Cell Signaling Technology) were incubated 1 hour at room temperature, and a 5 minute SuperSignal West Pico (Thermo Scientific, Waltham, MA, http://www.thermofisher.com/global/en/home.asp/) incubation followed. The film was developed, and band intensity was measured with Image J.
Ten trials of the vibrissae-evoked forelimb placing test was done on each side as described previously . Behavior was tested preoperatively for baseline performance and repeated weekly for 5 weeks post-dMCAo (blinded). Animals were excluded if their poststroke score did not decrease 10% or more from baseline.
Data are presented as means + SEM. Data were tested for normality and equal standard deviations in GraphPad InStat to determine the appropriate statistical test (parametric vs. nonparametric). The text and figure legends describe the statistical tests; unless stated differently, all tests were two-tailed. Differences were considered statistically significant at p < .05.
hCNS-SCns Enhance Functional Recovery and Prevent Secondary Cortical Damage in an hVEGF-Dependent Manner
hCNS-SCns were transplanted in the rat cortex 1 week poststroke and successful engraftment was observed (Fig. 1A, Supporting Information Figs. 1 and 7), consistent with our previous studies with these cells [26, 30]. Also consistent with our previous studies and that of others [31, 32], the cells migrated toward the lesion (Fig. 1A). The cells started migrating by 2 weeks post-transplantation but did not reach the lesion edge until 3–4 weeks post-transplantation.
The barrel field cortex, a somatosensory cortical region coding for whisker displacement, was consistently damaged by the lesion, therefore, a whisker stimulation test was employed to determine the effect of hCNS-SCns on functional recovery . Whisker-evoked paw reaching was significantly inhibited after stroke when the whiskers on the affected side (contralesional; Fig. 1B), but not the unaffected side (Supporting Information Fig. 2A), were stimulated. The affected whisker-paw response showed significantly greater recovery in cell-treated versus buffer-treated animals (Fig. 1B). Cell-induced recovery was significant as early as 1 week post-transplantation, which is before the cells have migrated to the lesion. A “dead hCNS-SCns” control group did not induce recovery (Supporting Information Fig. 2C, 2D) implying that viable cells were necessary for efficacy.
Expression of hVEGF by hCNS-SCns was confirmed in vitro by ELISA (1.5 ng/mg protein, minimum) and in vivo by immunofluorescent staining (Fig. 1E, Supporting Information Fig. 1). Avastin was injected to neutralize stem cell-secreted hVEGF in vivo and some Avastin staining was observed in the parenchyma around the graft and also associated with blood vessels in the penumbra (Supporting Information Fig. 3). Avastin is an anti-human VEGF antibody that does not bind rodent VEGF . Therefore, Avastin selectively inhibits the human VEGF secreted by the stem cells without affecting the host rodent VEGF. Avastin treatment significantly inhibited hCNS-SCns-enhanced functional recovery (Fig. 1C). Some inhibition was observed as early as 1 week post-treatment with Avastin; however, it was not significant until 3 weeks post-treatment (Fig. 1C). The Avastin effect was specific as treatment with a control antibody, IgG, had no effect and Avastin did not affect functional recovery in the buffer-treated animals (Supporting Information Fig. 2B). However, given the small degree of recovery in the buffer animals, Avastin-induced inhibition would be very small and possibly below the sensitivity of our assay.
The remaining ipsilesional cortex was significantly larger in the cell-treated group compared with the buffer group at 4 weeks post-transplantation (Fig. 1D), and a more healthy tissue architecture was apparent in the peri-infarct cortex of cell-treated animals (Supporting Information Fig. 4). Treatment with Avastin reduced this effect (Fig. 1D, Supporting Information Fig. 4). This implies that the cells reduce cortical atrophy that occurs at later stages after stroke, and this is dependent on hVEGF secretion.
hCNS-SCns Enhance BBB Integrity and Neovascularization After Stroke with Different Kinetics
BBB integrity, which is critical for proper brain function, is disrupted after stroke lasting up to several weeks [34, 35]. To determine the effect of hCNS-SCns on BBB permeability, a small molecule, biotin, was injected intravenously. hCNS-SCns-treated animals exhibited significantly less leakage of biotin into the peri-infarct parenchyma compared with buffer-treated animals (Fig. 2A, 2B), implying that hCNS-SCns enhance BBB integrity. This was an early host response to hCNS-SCns observed at 1 week post-transplantation; a similar trend was observed at 2 weeks post-transplantation.
To investigate the effect of hCNS-SCns on neovascularization, the density of lectin-perfused vessels (BVD) was analyzed in three cortical ROIs: the peri-infarct (ROI1), near the graft (ROI3), and between the two (ROI2; Fig. 2C). Cell-treated animals exhibited enhanced neovascularization after ischemia compared with buffer-treated animals (Fig. 2D, 2E); this observation was limited to the ipsilesional hemisphere with no effect in the contralesional side (Supporting Information Fig. 5A). hCNS-SCns-induced effects on vascularization were under spatial regulation and most pronounced in the peri-infarct region as opposed to ROIs closer to the graft (Fig. 2E). This implies that the peri-infarct microenvironment is responsive to hCNS-SCns, whereas the more healthy tissue is less permissive. However, a trend for smaller increases in vasculature that approached significance was observed in the other ROIs at later time points.
In addition to the spatial control, temporal regulation of hCNS-SCns-induced effects on vessel growth was observed. There was no observed effect on vessel density at 1 week post-transplantation when the cells elicit their effect on BBB integrity. The most pronounced effect of the cells in the peri-infarct cortex occurred later at 2 weeks post-transplantation and declined thereafter (Fig. 2E). Increased vascularization at 2 weeks was also confirmed by vessel length measurements (Supporting Information Fig. 5B). This timing mirrors the increase in vasculature observed in stroke-only animals, which peaks at 3 weeks post-stroke (equivalent to 2 weeks post-transplantation; slope coefficient week 1–week3 post-stroke = 0.16, p < .05; Supporting Information Fig. 5C).
Transplanted hCNS-SCns are very closely associated with vessels (Supporting Information Fig. 6), and thus ideally situated to influence the vasculature. However, they are distant from the peri-infarct cortex when they elicit their maximum effect on the vasculature (at 1 and 2 weeks post-transplantation), only migrating to this region at 3 and 4 weeks post-transplantation (Fig. 1). Therefore, hCNS-SCns function remotely to enhance peri-infarct vascular regeneration. This is in contrast to other reports that show transplanted cells act locally to enhance vascularization [11–13].
hVEGF Secretion Is Required for hCNS-SCns-Induced Neovascularization but Not BBB Repair
Avastin treatment significantly reduced hCNS-SCns-enhanced neovascularization in the peri-infarct cortex as determined both by vessel density (Fig. 3A, 3B) and vessel length analysis (Supporting Information Fig. 5B). This effect was not seen in buffer-treated animals confirming that Avastin did not block rodent VEGF. Furthermore, the IgG control had no effect on either hCNS-SCns- or buffer-treated groups, and Avastin did not adversely affect cell survival (Supporting Information Fig. 7B; percent cell survival 5.5 ± 1.5 [Avastin] vs. 6.9 ± 2.7 [IgG]). There was also no obvious effect of Avastin on hCNS-SCns differentiation; however, this was not quantifiable given the density of the graft at this time (Supporting Information Fig. 7C, 7D). Together, these data confirm that the effects of Avastin were specific to its neutralization of hVEGF. Thus, VEGF secretion by the transplanted cells is essential for increased vascularization. In contrast, Avastin had no effect on hCNS-SCns-induced BBB integrity (Fig. 3A, 3C) implying an hVEGF-independent effect of the cells on this parameter.
hCNS-SCns Enhance Angiogenic Signaling Pathways
Vascular regeneration and remodeling is regulated in part by signaling through the two main proangiogenic receptors VEGFR2 (Flk1) and Tie2, whose ligands are VEGF and Angiopoietin 1 and 2, respectively. Western blot analysis of cortical microvessels isolated at 1 week post-transplantation revealed that hCNS-SCns-treated animals exhibit enhanced phosphorylation of the VEGFR2 compared with the buffer group, indicative of increased signaling through the receptor (Fig. 4A, 4B: the same trend was observed with cell- and buffer-only treated animals, i.e., no antibody treatment, data not shown). Avastin blocked the cell-induced phosphorylation of VEGFR (with no effect in buffer-treated animals), implying a role of secreted hVEGF. Furthermore, hCNS-SCns also significantly enhanced expression of the angiopoietin receptor Tie2 at 1 week post-transplantation in cell-only- versus buffer-only- treated groups (buffer vs. cell [band intensity]: 0.97 ± 0.2 vs. 3.1 ± 0.6, p < .01, Supporting Information Fig. 8). The same trend was observed in cell and buffer IgG-treated animals and blocked by Avastin (Fig. 4A, 4B); however, statistical significance was not reached in either case.
hCNS-SCns Enhance Tight Junction and βDG Protein Expression After Stroke
Tight junctions (TJs) between endothelial cells are critical for BBB integrity and are formed by TJ proteins [36, 37]. Western blot analysis of microvessels isolated from the ischemic cortex of hCNS-SCns-treated animals at 1 week post-transplantation revealed significant increased expression of the TJ proteins occludin and claudin-5, but no change in ZO-1 expression compared with the buffer group (Fig. 4C, 4D: similarly observed with cell-only- and buffer-only-treated animals, data not shown). This is consistent with increased BBB integrity in cell-treated animals at this time. Surprisingly, the addition of Avastin significantly reduced the hCNS-SCns-induced expression of occludin and claudin-5 in cell-treated animals (Fig. 4C, 4D) without an apparent change in BBB leakage (Fig. 3C); however, occludin expression did remain significantly higher in cell/Avastin-treated versus buffer/Avastin-treated animals (p < .05 cell/Avastin vs. buffer/Avastin).
Astrocytic endfeet surrounding the vessels (Fig. 5A) also contribute to BBB integrity [36, 38], and dystroglycan (DG), which binds astrocytes to the endothelial cell extracellular matrix, may be involved [39–41]. DG appears to be associated with all microvessels in the cortex (Supporting Information Fig. 9) and at least in larger vessels, colocalized with the astrocytic endfoot marker aquaporin-4 rather than endothelial cells (Fig. 5B, 5C). hCNS-SCns enhanced expression of the β subunit of DG (βDG) after stroke (Fig. 5D). However, this was not observed at 1 week post-transplantation when the cells elicit their effect on BBB integrity, indicating that enhanced βDG expression is not involved in the initial cell-induced repair of the BBB. βDG expression exhibited a similar spatiotemporal pattern as observed for lectin staining of blood vessels (compare Figs. 2E and 5D), with the most significant increase in expression in the peri-infarct cortex (ROI 1 Fig. 5D) starting at 2 weeks post-transplantation. Furthermore, Avastin blocked the cell-induced expression of βDG (Fig. 5E), implying a role of hVEGF in this effect.
hCNS-SCns Decrease the Inflammatory Response After Stroke
hCNS-SCns-treated animals had significantly fewer Iba1-positive monocytes/macrophages in the peri-infarct cortex than buffer-treated animals at 1 week post-transplantation and a similar trend was measured at 2 weeks post-transplantation (Fig. 6A, 6B); no significant difference in Iba1-positive cells between buffer- and hCNS-SCns-treated animals was observed in the rest of the cortex (ROI2). hCNS-SCns treatment also reduced the number of Iba1-positive cells in IgG-injected animals compared with buffer/IgG-treated animals (Fig. 6C); this immunomodulation was blocked by Avastin treatment, implying that hVEGF is important for this immunomodulatory effect. The cells could not fully suppress the inflammatory response induced by adding IgG (compare cell/IgG with cell-only); this is likely because the inflammatory response is too great. However, the fact that the cells had a measurable suppressive effect even in the presence of such a large inflammatory response implies they have significant immunomodulatory properties.
Correlating hCNS-SCns-Induced Changes in the Brain with Functional Recovery
We compared the timing of hCNS-SCns-induced behavior recovery to cell-induced changes in the brain and found that, although vascularization is postulated to be important for recovery after stroke, significant recovery was observed before effects on the vasculature were seen (Fig. 7A). This early phase of recovery (at 2 weeks poststroke, i.e., 1 week post-transplantation) correlated with hCNS-SCns-induced changes in inflammation and BBB integrity. This raises the question whether hCNS-SCns-enhanced recovery is based solely on these early events (i.e., BBB and inflammation changes) and whether neovascularization is involved at all. However, an interesting observation about the pattern of recovery hints that neovascularization may be important for recovery at later time points as examination of all the individual cell-treated animals revealed two distinct groups with respect to recovery patterns. Group I exhibited early recovery within the first week of transplantation before vascularization; Group II exhibited delayed recovery starting in the third week post-transplantation coincident with neovascularization (Fig. 7B). In contrast, buffer-treated animals exhibited a more random recovery pattern commencing at all time points poststroke (data not shown). Furthermore, Avastin, which inhibits hCNS-SCns-induced vascularization, clearly inhibits recovery in Group II (compare group II, Fig. 7B, 7C).
Taken together, our data demonstrate that hCNS-SCns influence multiple parameters of brain repair after stroke with different temporal profiles as summarized in the schematic (Fig. 7D), and that these effects are mediated, at least in part, by secretion of hVEGF.
We show that the functional recovery observed after injection of hCNS-SCns cells in vivo was a direct effect of the hVEGF secreted by the cells, as Avastin, which neutralizes human but not rodent VEGF, inhibited the functional improvements. Others have reported that overexpression of VEGF in transplanted cells further enhances recovery [7, 8]. However, in these studies it is unclear if VEGF acts directly on the host brain or indirectly by enhancing graft survival. Thus, such studies do not distinguish whether VEGF is necessary for recovery or alternatively amplifies recovery induced by some other mechanism. This study therefore provides the first direct evidence of a stem cell-secreted factor that is critical for cell-induced functional recovery in the poststroke brain.
Having established a critical role for secreted hVEGF in functional recovery, we then investigated what changes hCNS-SCns elicit in the brain to enhance recovery and in particular, which of these changes are modulated by hVEGF. We and others [11–13, 42] have postulated a role for stem cell-induced vascularization in cell-induced recovery; this fits with the essential role of hVEGF in recovery as we show the first in vivo evidence that the induced vascularization is hVEGF-dependent. We also demonstrate for the first time that hCNS-SCns-induced neovascularization is regulated in a spatiotemporal manner. The most significant increase was observed in the peri-infarct region at 2 weeks post-transplantation, with very little effect in areas closer to the graft. This implies an effect that is not simply a uniform response to a gradient of graft-secreted factors, but one that requires host tissue responsiveness to the graft. Therapeutically, this is important as it shows hCNS-SCns do not readily affect vascularization in healthy tissue but primarily influence tissue already undergoing repair and revascularization. Furthermore, hCNS-SCns-treated animals demonstrate a profile of vessel induction followed by regression suggesting a highly regulated process rather than a continuous, aberrant, and potentially detrimental increase in blood vessel formation as is observed in certain retinopathies.
The peri-infarct tissue is presumably the most responsive region to hCNS-SCns as this area upregulates angiogenic signaling after ischemia through the VEGFR2 and the angiopoietin receptors Tie 1 and 2 . We found that hCNS-SCns treatment increases endothelial cell signaling through VEGFR2 and increases Tie2 expression in vivo prior to the observed changes in vascularization. Furthermore, we demonstrated for the first time in vivo that secretion of hVEGF by hCNS-SCns is both necessary and sufficient for the increased angiogenic signaling as it was abolished by Avastin. In addition to changes in the blood vessel number, we found that hCNS-SCns also enhances expression of βDG in a hVEGF-dependent manner. βDG is an extracellular matrix adhesion protein abundant in astrocytic endfeet surrounding blood vessels that may also be expressed by pericytes and endothelial cells [39, 44]. Thus, this data implies that the hCNS-SCns affect not only endothelial cells but also other critical components of the vasculature, which may be indicative of enhanced communication and functioning of the neurovascular unit .
The exact contribution of enhanced neovascularization to stem cell-induced functional recovery is however unclear as VEGF is known to have pleiotropic effects, including influencing neurite outgrowth and neurogenesis (reviewed in ), all of which could contribute to hVEGF-induced functional recovery. What is apparent from our study is that neovascularization is unlikely to be involved in the initial phase of recovery, as hCNS-SCns significantly enhanced behavior recovery at 1 week post-transplantation, which is prior to their effects on vessel formation. Omori et al.  also concluded that angiogenesis was not the only contributing factor for human mesenchymal stem cell (hMSC)-induced functional recovery, as different hMSC treatments resulted in different levels of functional recovery but had similar effects on angiogenesis.
The early cell-induced recovery at 1 week post-transplantation instead coincided with hCNS-SCns effects on inflammation and BBB repair. Previous studies have reported on the unexpected anti-inflammatory effects of neural stem cells or MSCs and their ability to improve BBB integrity [48, 49] when administered either before or acutely (6–72 hours) after stroke [20–23]. Our study expands this therapeutic window and demonstrates that subacute delivery of hCNS-SCns remains immunosuppressive and can restore impaired BBB function. Although the T cell deficiency of our Nude rat model could potentially bias the inflammatory data, this situation is not unlike that of the patient as they will be given immunosuppressive drugs to inhibit T cells to prevent neural graft rejection.
It is not understood how xenografts of human stem cells into rodents are immunosuppressive; our data demonstrates for the first time that secretion of hVEGF plays an important role in this immunomodulatory effect of hCNS-SCns. This was unexpected as VEGF is conventionally considered to be proinflammatory. However, a growing body of literature corroborates the anti-inflammatory properties of VEGF. Manoonkitiwongsa et al.  found that treatment of ischemic brains with low doses of VEGF reduced macrophage numbers, whereas higher doses increased macrophage density. Furthermore, hematopoietic progenitor cells express VEGF receptors and VEGF has been shown to directly inhibit the growth of myeloid progenitor cells (the monocyte and granulocyte precursors) . Additionally, VEGF reportedly suppresses the development and activation of dendritic cells (antigen-presenting cells important in initiating immune responses) and T cells [52, 53]. Preclinical and clinical studies indicate that the immunosuppressive effect of VEGF may be involved in establishing immune privilege of VEGF-secreting tumors . Thus, precedence exists for an immunosuppressive role for VEGF, which may be dependent on the dose, timing, and route of VEGF delivery, suggesting the importance of further investigation of VEGF in the context of cerebral ischemia and cell transplantation.
The immunosuppressive effects of hCNS-SCns may contribute to hCNS-SCns-induced BBB repair as inflammation contributes to BBB breakdown [17, 18, 37], and there is significant crosstalk between inflammatory cells and components of the neurovascular unit . Additionally, hCNS-SCns enhanced other effectors of the BBB including increased expression of Tie2 and TJ proteins. A similar result was reported by Zacharek et al.  after acute delivery of MSCs, implying this effect is independent of the cell type transplanted. Avastin blocked both hCNS-SCns-induced immunomodulation and TJ expression; these may be interrelated as inflammation is known to affect TJ proteins expression . However, although hVEGF secretion was important for these aforementioned parameters, cell-enhanced BBB integrity appeared to be hVEGF-independent as it was not blocked by Avastin. A possible explanation for this disparity is that occludin expression, although reduced in the presence of Avastin, still remained significantly elevated in cell-treated animals compared with the buffer control group and perhaps retaining a certain threshold level of occludin may be enough for sustained BBB integrity. It remains to be determined whether transcellular influx (i.e., through the endothelial cells) of biotin through its known transporter  or by pinocytosis is also affected in our experimental paradigm, as such mechanisms are less influenced by TJs than paracellular leakage (i.e., between endothelial cells).
Understanding how cell transplantation affects the host brain will lead to further improvements in cell efficacy and, perhaps more importantly, may highlight potential side effects of cell therapy. Furthermore, cell-induced changes in the brain could serve as useful surrogate clinical indicators of transplanted cell activity. For example, changes in vasculature and blood flow can be measured noninvasively in patients by perfusion studies. Moving from the bench to the clinic also raises cell manufacturing issues, in particular designing bioassays to predict clinical efficacy of the cell product . Understanding the mechanism of action of the transplanted cells will be critical for this, and our data implies that secreted VEGF could be an important predictive marker of efficacy, especially as cell viability may affect the capacity of stem cells to secrete VEGF and other potential biofactors. However, future studies are required to determine the relationship between the amount of hVEGF secreted and the extent of cell-induced recovery.
In summary, we established that stem cell-secreted VEGF is essential for hCNS-SCns-induced recovery; no doubt other secreted factors will also be involved. Furthermore, functional recovery correlated with hCNS-SCns-induced suppression of inflammation, increased vessel formation, and enhanced BBB integrity; hVEGF was important for the immunosuppressive and neovsacularization effects of the hCNS-SCns, and may possibly be involved in altering BBB integrity. Thus, subacute cell transplantation therapy offers a multimodal strategy for brain repair that could significantly expand the therapeutic window for stroke. Finally, as vascular degeneration and inflammation are hallmarks of many other cerebral pathologies  including meningitis, multiple sclerosis, and vascular dementia, further research focused on the vascular repair and immunosuppressive properties of transplanted cells could have far reaching implications as stem cell therapies are developed and advanced to the clinic.
We thank Calvin Kuo, Richard Daneman, Carolina Maier, Pak Chan, Marion Buckwalter, Katrin Andreasson, and Napoleone Ferrara for helpful discussions. We also thank Purnima Narasimhan and Dongping He for technical assistance; Alexandra Capela, Cindy Samos, and Bruce Schaar for comments on the article; and Beth Hoyte for help with the figures. This work was supported by National Institutes of Health National Institute of Neurological Disorders and Stroke (NS058784, NS27292, NS37520 to G.K.S.), the William Randolph Hearst Foundation, Bernard and Ronni Lacroute, Russell and Elizabeth Siegelman, and the Edward E. Hills Fund (to G.K.S.). M.P.P. is a recipient of a postdoctoral fellowship from the Ministry of Education and Science of Spain (reference 2007-1219). N.H. is currently affiliated with the Department of Neurosurgery, Nagasaki University School of Medicine, Sakamoto, Nagasaki, Japan.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
S.H. is a full time employee of StemCells Inc., the company that provided the cells and has equity (stock) in the company. The other authors have no potential conflict of interest.