OUP user menu

Anti-inflammatory mechanism of intravascular neural stem cell transplantation in haemorrhagic stroke

Soon-Tae Lee, Kon Chu, Keun-Hwa Jung, Se-Jeong Kim, Dong-Hyun Kim, Kyung-Mook Kang, Nan Hyung Hong, Jin-Hee Kim, Jae-Joon Ban, Hee-Kwon Park, Seung U. Kim, Chung-Gyu Park, Sang Kun Lee, Manho Kim, Jae-Kyu Roh
DOI: http://dx.doi.org/10.1093/brain/awm306 616-629 First published online: 20 December 2007


Neural stem cell (NSC) transplantation has been investigated as a means to reconstitute the damaged brain after stroke. In this study, however, we investigated the effect on acute cerebral and peripheral inflammation after intracerebral haemorrhage (ICH). NSCs (H1 clone) from fetal human brain were injected intravenously (NSCs-iv, 5 million cells) or intracerebrally (NSCs-ic, 1 million cells) at 2 or 24 h after collagenase-induced ICH in a rat model. Only NSCs-iv-2 h resulted in fewer initial neurologic deteriorations and reduced brain oedema formation, inflammatory infiltrations (OX-42, myeloperoxidase) and apoptosis (activated caspase-3, TUNEL) compared to the vehicle-injected control animals. Rat neurosphere-iv-2 h, but not human fibroblast-iv-2 h, also reduced the brain oedema and the initial neurologic deficits. Human NSCs-iv-2 h also attenuated both cerebral and splenic activations of tumour necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), and nuclear factor-kappa B (NF-κB). However, we observed only a few stem cells in brain sections of the NSCs-iv-2 h group; in the main, they were detected in marginal zone of spleens. To investigate whether NSCs interact with spleen to reduce cerebral inflammation, we performed a splenectomy prior to ICH induction, which eliminated the effect of NSCs-iv-2 h transplantation on brain water content and inflammatory infiltrations. NSCs also inhibited in vitro macrophage activations after lipopolysaccharide stimulation in a cell-to-cell contact dependent manner. In summary, early intravenous NSC injection displayed anti-inflammatory functionality that promoted neuroprotection, mainly by interrupting splenic inflammatory responses after ICH.

  • neural stem cell
  • spleen
  • cerebral inflammation
  • intracerebral haemorrhage
  • macrophage


Neural stem cell (NSC) transplantation has been studied as a promising tool for inducing regeneration of damaged brain tissue in many neurological disorders (Miller, 2006). We have shown that the intravenous transplantation of human NSCs promotes neurological recovery with cell replacement in intracerebral haemorrhage (ICH) (Jeong et al., 2003). However, recent evidence indicates that NSC transplantation may protect the CNS from inflammatory damage via a ‘bystander’ mechanism rather than by direct cell replacement (Martino and Pluchino, 2006). NSCs appear to rescue degenerating neurons by modulating the host environment by adopting a chaperone-like role (Ourednik et al., 2002; Hagan et al., 2003). In experimental autoimmune encephalomyelitis (EAE) models, NSCs exert immune-like functions and induce the apoptosis of blood-borne encephalitogenic T cells (Pluchino et al., 2003, 2005), as well as decrease CNS inflammation via peripheral suppression of the adaptive immune response (Einstein et al., 2007). This indicates the feasibility of using NSCs to reduce cerebral inflammation and secondary brain damage after stroke. However, no previous study has been performed on the possible anti-inflammatory potential of NSCs on innate inflammatory response after acute stroke.

Intracerebral haemorrhage (ICH) represents at least 15–20% of all strokes in the Caucasian population (Qureshi et al., 2001), and occurs with increasing frequency as a complication of the thrombolytic treatment of ischaemic stroke (Lapchak, 2002). ICH induces neurological damage due to local tissue deformation and the subsequent developments of excitotoxicity, apoptosis, and inflammation (Aronowski and Hall, 2005). Of these, ICH-induced inflammation appears to be a key factor of secondary brain damage, which suggests that anti-inflammatory approaches may lessen the outcome of haemorrhagic stroke (Jung et al., 2004; Chu et al., 2004a; Aronowski and Hall, 2005; Wang and Dore, 2007; Sinn et al., 2007).

Both innate and adaptive inflammatory mechanisms have been clearly shown to participate in the progression of cellular injury after stroke (Hallenbeck et al., 2005). A prominent inflammatory response, which includes the activation of resident brain microglia and inflammatory infiltration into the brain, is initiated by neutrophils followed by macrophages (Del Bigio et al., 1996). This process is accompanied by the massive, rapid activation of the peripheral immune system and the dynamic and widespread activation of inflammatory cytokines and chemokines in spleen (Offner et al., 2006). In particular, the spleen is the key organ that activates tumour necrosis factor-alpha (TNF-α) and nuclear factor-kappa B (NF-κB) in systemic inflammatory response (Huston et al., 2006; Tracey, 2007). TNF-α is one of the major inflammatory mediators in stroke (Zheng and Yenari, 2004), and is largely produced by infiltrating macrophages (Gregersen et al., 2000). Thus, the issue as to whether NSCs can modulate cerebral and peripheral inflammatory response in acute stroke has important therapeutic relevance that should be clarified before developing NSC transplantation strategies.

In this study, we examined the influence of systemic NSC transplantation on brain and spleen inflammatory reactions during the acute period of ICH, and we observed that NSCs have an important ‘bystander’ anti-inflammatory effect on the spleen-macrophage system.

Material and Methods

Human neural stem cell culture

All experimental procedures were approved by the Care of Experimental Animals Committee at Seoul National University Hospital and human cell use was approved by our institutional review board. Primary dissociated neural stem cells were prepared from the ventricular zone (VZ) of fetal human brain (15 weeks gestation) and maintained. Detailed characteristics of this human NSC line (H1 clone) have been described elsewhere (Ourednik et al., 2002; Jeong et al., 2003; Chu et al., 2004b,c,d; Imitola et al., 2004; Chu et al., 2005; Lee et al., 2005; Parker et al., 2005; Lee et al., 2006a). In brief, suspensions of dissociated NSCs (5 × 105 cells per ml), isolated from the telencephalic VZ of a 15-week human fetal cadaver, were cultured in 98% DMEM (Dulbecco's Modified Eagle's Medium)/F12 (GIBCO, Invitrogen, Carlsbad, CA, USA), 1% N2 supplement (GIBCO), 1% penicillin/streptomycin (GIBCO), 8 mg/ml heparin (Sigma), 10 ng/ml leukemia inhibiting factor (Chemicon-Millipore), 20 ng/ml basic fibroblast growth factor (bFGF, Calbiochem) in uncoated 25 cm2 flasks (Falcon™, BD Bioscience) at 37°C in 5% CO2. NSCs were cultured in poly-L-lysine-coated culture dishes or flasks as single cells or large clusters, which can be subcultured and passaged weekly over a period of 6 months. This cell line can give rise to multiple neural cell types throughout the neural axis during development and has the ability to reconstitute these regions when perturbed (Snyder et al., 1997; Ourednik et al., 2001; Parker et al., 2005). In addition, NSC clones generated using the same method expresses all commonly accepted NSC markers (nestin, musashi, vimentin, etc.) and can form neurospheres (Kitchens et al., 1994; Parker et al., 2005). When injected into rats for histologic analysis, NSCs were pre-labelled with Vybrant® DiO cell-labelling solution (Invitrogen) before transplantation, according to the manufacturer's instructions.

Rat neurosphere and human fibroblast culture

We cultured neurospheres using the published method (Singec et al., 2006), on postnatal days 5–7 from Sprague-Dawley rats (Orient Bio, South Korea). We used microdissected sections containing the subventricular zone (SVZ) for the culture, and used DMEM/F12 medium (GIBCO) supplemented with B27 (Invitrogen), 20 ng/ml bFGF (PeproTech, Rocky Hill, NJ, USA), and 20 ng/ml epidermal growth factor (EGF, PeproTech) for the culture medium. For passaging, we collected neurospheres with 37°C warmed 0.05% trypsin, centrifuged them, and mechanically triturated them, and we plated single cells again after counting cells. For the transplantation, neurospheres (at three passages) were dissociated into single cells, and suspensions containing 5 × 106 cells in 0.5 ml PBS were used. Primary human fibroblasts were cultured from a normal skin biopsy specimen with the patient's consent using the published method (Lee et al., 2006c). Cell suspensions containing 5 × 106 fibroblasts in 0.5 ml PBS were used for transplantation.

Induction of ICH and human NSC transplantation

One hundred and seventy nine male Sprague-Dawley rats (Orient Bio), each weighing 210–240 g, were utilized. Experimental ICH was induced via the stereotaxic intrastriatal administration of bacterial collagenase type VII (0.23 IU dissolved in 1 μl saline, Sigma), as described previously (Jeong et al., 2003; Chu et al., 2004a; Jung et al., 2004; Lee et al., 2006b). Sham ICH was induced with a stereotaxic needle insertion and an injection of equal volume (1 μl) of saline instead of collagenase. Two or 24 h after ICH induction, human NSCs were administered intravenously (via tail vein, 5 million cells in 0.5 ml PBS/rat) or intracerebrally [1 million cells in 2 μl PBS/rat, divided at three co-ordinates (anterior–posterior, medial–lateral, dorsal–ventral): (+1 mm, +1.4 mm, −3.0 mm), (+0.2 mm, +2.0 mm, −3.0 mm), (−0.4 mm, +2.2 mm, −3.0 mm), respectively], as previously described (Jeong et al., 2003; Kelly et al., 2004), namely intravenous transplantation at 2 h (the NSCs-iv-2 h group), intravenous transplantation at 24 h (the NSCs-iv-24 h group), intracerebral transplantation at 2 h (the NSCs-ic-2 h group), or intracerebral transplantation at 24 h (the NSCs-ic-24 h group). Vehicle was administered to the ICH-vehicle group by injecting 0.5 ml of PBS via tail vein at 2 h after the ICH induction.


Rats were anesthetized with ketamine and xylazine (Sigma). Spleens were identified by midline laparotomy and removed after appropriate blood vessel ligation.

Behavioural testing

Behavioural testing (n = 10 for each group) was conducted weekly for 5 weeks after ICH using modified limb placing tests (MLPT), as previously described (Jeong et al., 2003; Chu et al., 2004a; Lee et al., 2006b). This test consists of two limb-placing tasks, which assess the sensorimotor integration of the forelimb and the hindlimb, by monitoring the subject's responses to tactile and proprioceptive stimulation. First, the rat is suspended 10 cm above a table, and the stretch of the forelimbs towards the table is observed and evaluated: a normal stretch is scored as 0 points; abnormal flexion is scored as 1 point. Next, the rat is positioned along the edge of the table, with its forelimbs suspended over the edge, and is then allowed to move freely. Each forelimb (forelimb-second task, hindlimb-third task) is gently pulled down, and retrieval and placement is evaluated. Finally, the rat is placed near the table edge, in order to assess the lateral placement of the forelimb. The three tasks are scored in the following manner: normal performance is scored as 0 points; delayed (at least 2 s) and/or incomplete performance is scored as 1 point; no performance is scored as 2 points. A total score 7 points indicates maximal neurological deficit, and a score of 0 points denotes normal performance.

Spectrophotometric assay of haemorrhage volume

Haemorrhage volumes were quantified at 3 days post-ICH using a spectrophotometric assay, as described previously (n = 6 for each group) (Choudhri et al., 1997; Asahi et al., 2000; Lee et al., 2006b). In brief, hemispheric brain tissue was acquired from normal rats who had been subjected to complete transcardial perfusion for the removal of intravascular blood. Incremental volumes of homologous blood (0–200 μl) were added to each hemispheric sample, along with PBS, to a total volume of 3 ml, followed by 30 s of homogenization, 1 min of sonication on ice, and 30 min of centrifugation at 12 000 r.p.m. Drabkin's reagent (1.6 ml; Sigma) was then added to 0.4 ml aliquots, and allowed to stand for 15 min at room temperature. Optical density (OD) was then measured and recorded at 540 nm with a spectrophotometer (Molecular Devices). These procedures showed a linear relationship between the haemoglobin concentrations in the perfused rat brain and the volume of added blood. Measurements from the perfused ICH brains were then compared with this standard curve, allowing us to obtain data regarding haemorrhage volume (in μl).

Measurements of brain water content

Water contents were measured at 3 days after ICH (n = 6 for each group), as previously described (Jeong et al., 2003; Chu et al., 2004a; Jung et al., 2004; Lee et al., 2006b). In brief, the rats were anesthetized and sacrificed via decapitation. The rats’ brains were removed immediately, and divided into two hemispheres along the midline, after which the cerebellum was removed from each brain. The brain samples were then immediately weighed on an electronic analytical balance to obtain wet weights, then dried in a gravity oven at 100°C for 24 h, in order to obtain dry weights. Water contents were expressed as a percentage of wet weight: the formula used to calculate the water contents was as follows: (wet weight − dry weight)/(wet weight).

Tissue preparation and immunohistochemistry

At 3 days or 35 days after ICH, rats (n = 4 in each of the groups) were re-anesthetized and perfused through the heart with 50 ml of cold saline and 50 ml of 4% paraformaldehyde, in 0.1 mol/l PBS. Sections (30 μm thickness) of brains were immunostained with antibodies against myeloperoxidase (MPO; DAKO), Ox-42 (Chemicon), or activated caspase-3 (BD biosciences), and sections of spleens were stained with antibodies against ED1 (Serotec, Oxford, UK), CD11b (Chemicon), choline acetyltransferase (ChAT, Serotec), and human nuclear antigen (HuN, Chemicon), as previously described (Jeong et al., 2003; Chu et al., 2004a; Jung et al., 2004; Lee et al., 2006b). The terminal deoxynucleotidyl transferase biotin-dUTP nick end labelling (TUNEL) was used to detect in situ DNA fragmentation using a commercial kit (R&D systems). For fluorescent double staining of TUNEL and cell markers, we used anti-NeuN (Chemicon), anti-glial fibrillary acidic protein (GFAP, Chemicon) antibodies, and a fluorescent TUNEL assay kit (Chemicon). Fluoro-Jade C staining was performed according to the published protocol to detect neurons undergoing degeneration (Schmued et al., 2005). To confirm any cell-to-cell contact or co-labelled cell markers, sections were analyzed with a laser-scanning confocal microscopic (LSM 510, Carl Zeiss microimaging, Thornwood, NY, USA) imaging system.

One coronal brain section taken through centers of needle tracts and two adjacent sections (1 mm width) were analysed by counting marker-specific cells throughout entire sections (three sections per antibody staining). Total counts were converted to cell densities in order to facilitate quantification. We investigated NSC distributions in brain, spleen, lungs, livers and mesenteric lymph nodes of the NSCs-iv-2 h groups with DAPI staining at 3 or 35 days after ICH. ED1+-activated macrophages in the spleen were counted in three marginal zone areas immediately adjacent to the white pulp in at least three sections per animal using a magnification of 400× (high power field, HPF) over a microscopic field. For determining hemispheric atrophy, three sections through the needle entry site were Nissl stained at 35 days after ICH (n = 6 per group). The total hemispheric areas of each section were traced and measured using image analyzer (Image Pro-Plus, Media Cybernetics, Bethesda, MD, USA) as described previously (Matsushita et al. 2000; Chu et al. 2004a; Jung et al. 2004). The hemispheric atrophy was expressed as a percentage of contralateral hemispheric area.

Western blotting and RT-PCR

Twenty four hours after ICH induction, rats were sacrificed by decapitation, and brains, spleens and mesenteric lymph nodes were immediately extracted (n = 4 in each group). Homogenates of ipsilateral (haemorrhagic) hemispheres and spleens were serially processed for western blotting, as described previously (Jung et al., 2004; Lee et al., 2006d), using anti-NF-κB (1 : 200; BD Biosciences) and anti-β-actin antibody (Santa Cruz Biotechnology). Relative optical densities were calculated versus measured values of β-actin. We also conducted RT-PCR for TNF-α, interleukin (IL)-1β, IL-4, IL-6, tumour growth factor (TGF)-β1, Fas and Fas ligand (FasL) at 24 h after ICH induction (n = 3 in each group), as previously described (Lee et al., 2006d,). In brief, total RNA was isolated from homogenates (50 mg) of haemorrhagic hemispheres, spleens, or mesenteric lymph nodes using TRI reagent (Sigma). RT-PCRs were conducted using First strand cDNA Synthesis kits (Roche, Basel, Switzerland) over 25 cycles of 95°C, 58°C and 72°C for 40 s each using the following primer sets: TNF-α: 5′-TAC TGA ACT TCG GGG TGA TTG GTC C-3′ (sense) and 5′-CAG CCT TGT CCC TTG AAG AGA ACC-3′ (antisense); IL-1β: 5′-ATG GCA ACT GTC CCT GAA CT (sense) and 5′-GTC ATC ATC CCA CGA GTC AC (antisense); IL-4: 5′-CTG CTT TCT CAT ATG TAC CGGG (sense) and 5′-TTT CAG TGT TGT GAG CGT GG (antisense); IL-6: 5′-CTT GGG ACT GAT GTT GTT GAC-3′ (sense) and 5′-CTC TGA ATG ACT CTG GCT TTG-3′ (antisense); TGF-β1: 5′-CTA ATG GTG GAC CGC AAC AAC-3′ (sense) and 5′-CGG TTC ATG TCA TGG ATG GTG-3′ (antisense); Fas: 5′-CAA GGG ACT GAT AGC ATC TTT GAG G-3′ (sense) and 5′-TCC AGA TTC AGG GTC ACA GGT TG-3′ (antisense); Fas-L: 5′-CAG CCC CTG AAT TAC CCA TGT C-3′ (sense) and 5′-CAC TCC AGA GAT CAA AGC AGT TC-3′ (antisense). Transcript levels were quantified by analyzing the scanned photographs of gels using appropriate imaging software (Molecular Analyst®, Bio-Rad). Relative optical densities were determined by comparing measured values with GAPDH mean values.

FACS analysis

At 3 days after the ICH induction, single cell suspensions (5 × 105 cells) were prepared from spleens in phosphate buffered saline (PBS) supplemented with 2% fetal calf serum (FCS) and 0.05% sodium azide. To remove contaminating erythrocytes, the cell suspensions were subjected to osmotic lysis using a hypotonic ammonium chloride solution. With using FITC anti-rat CD11b (2 μg/ml, BD Biosciences) and Phycoerythrin (PE) anti-rat TNF-α (2 μg/ml, eBioscience, San Diego, CA, USA) antibodies, 5 × 104 cells for each sample were analyzed on a FACS II flow cytometer (BD Bioscience).

Co-culture of rat macrophages with NSCs

Peritoneal macrophages were isolated from Sprague-Dawley rats, as previously described (Falcone and Ferenc, 1998). Briefly, rats were injected i.p. with 3 ml of 4% thioglycollate. Four days later cells were harvested by lavage with cold PBS. Peritoneal cells were recovered by centrifugation and resuspended in 24-well plates in 500 μl of RPMI-10% FBS containing penicillin (200 U/ml) and streptomycin (0.2 mg/ml). Cells were allowed to adhere for 2 h and then washed free of non-adherent cells. Adherent cells were then re-seeded (106 cells per well) in 500 μl of RPMI. Two hours before or after lipopolysaccharide (LPS) stimulation, NSCs (106 cells per well) were added to these macrophages as mixed culture. For the double chamber system, equal numbers of NSCs were plated into transwell membrane inserts (Costar) with a 0.4 μm pore size to avoid direct cell-to-cell interactions and to allow the diffusion of soluble factors between two cell populations (Shen et al., 2004). LPS was added to the cultures at a final concentration of 100 ng/ml. Supernatants were collected 4 or 24 h after adding LPS and prepared for TNF-α and IL-6 ELISA using ELISA kits (R&D Systems).

Statistical analysis

All data in this study are presented as means ± SD. Error bars in the figures mean SD except Fig 1, in which the error bars are SE. The Mann–Whitney U-test was used for inter-group comparisons and we did not specify the test in these cases. For a nonparametric comparison among three and more unpaired groups, we used Kruskall-Wallis analysis of variances (ANOVA) and specified the test. When P-values from Kruskall–Wallis was <0.05, Mann–Whitney U-test was further used for post-hoc inter-group comparisons. A two-tailed probability value of <0.05 was considered to be significant.


NSCs-iv-2 h reduced initial neurologic deterioration

We previously reported favourable neurologic recovery in the intravenous NSCs transplantation at 24 h after the ICH induction (NSCs-iv-24 h group) compared to the vehicle-treated ICH group (ICH-vehicle) (Jeong et al., 2003). Thus, we first investigated whether an earlier NSCs transplantation at 2 h after ICH (NSCs-iv-2 h) can induce more favourable results. Although the ICH-vehicle group scored over six points on the MLPT at 1 day after ICH (6.85 ± 0.38), the NSCs-iv-2 h group exhibited less profound initial deficits (4.66 ± 0.98, P < 0.01, Fig. 1). The NSCs-iv-2 h group then continued to recover, and the difference between the two groups was statistically significant until at least 5 weeks after ICH (all P < 0.01). Although the MLPT score of NSCs-iv-24 h group started to differ from that of the ICH-vehicle group at 4 weeks, the NSCs-iv-2 h group showed less neurologic deficits than the NSCs-iv-24 h group from day 1 to day 35 (all P < 0.01). We also investigated the neurologic deficits of the NSCs-ic-2 h and NSCs-ic-24 h groups for comparison. Although the MLPT scores of these two groups started to differ from that of the ICH-vehicle group at 21 or 28 days after the ICH, the neurologic deficits during the early periods (day 1–14) were not different to those of the ICH-vehicle group (Fig. 1). In summary, only the NSCs-iv-2 h group showed less initial neurologic deficits compared to the other groups.

Fig. 1

Modified limb placing test (MLPT). At 24 h after ICH, the NSCs-iv-2 h group exhibited less profound deficits than the ICH-vehicle group, and then progressively recovered. The MLPT score of NSCs-iv-24 h group started to differ from that of the ICH-vehicle group at 4 weeks after ICH. NSCs-iv-2 h group showed less neurologic deficits than NSCs-iv-24 h group from day 1 to day 35. The MLPT scores of the NSCs-ic-2 h and the NSCs-ic-24 h groups started to differ from that of the ICH-vehicle group at 21 or 28 days respectively, and the neurologic deficits during the early periods (day 1–14) were not different to those of the ICH-vehicle group. *P < 0.05, **P < 0.01 versus ICH-vehicle. P < 0.01 versus NSCs-iv-24 h. P < 0.05, §P < 0.01 versus NSCs-ic-2 h.

NSCs-iv-2 h reduced brain water content

Because the lower levels of neurologic deficits observed in the NSCs-iv-2 h group at 1 day after ICH could not be explained by cell replacement or regeneration, we hypothesized that some form of initial neuroprotection or anti-inflammation had been initiated in the NSCs-iv-2 h group. Thus, we measured brain water contents at 3 days after ICH, because water content represents brain oedema which is one of the most important surrogate markers of perihematomal inflammation and tissue damage and peaks at 3–4 days after ICH (Xi et al., 2006). ANOVA suggested an intergroup difference among the ipsilateral brain water contents of the five groups (p < 0.05, Kruskall–Wallis ANOVA, Fig. 2A). Mean brain water content in ipsilateral (haemorrhagic) hemispheres was 81.63 ± 0.55% in the ICH-vehicle group, and 80.10 ± 0.27% in the NSCs-iv-2 h group (P < 0.01), whereas in contralateral (non-haemorrhagic) hemispheres these were 79.63 ± 0.36% in the ICH-vehicle group and 79.30 ± 0.25% in the NSCs-iv-2 h group (P = 0.08), demonstrating that NSCs-iv-2 h significantly reduced brain oedema in haemorrhagic hemispheres. However, the brain water contents in the other groups, i.e. the NSCs-iv-24 h group [81.18 ± 0.45% (ipsilateral); 79.58 ± 0.26% (contralateral)], the NSCs-ic-2 h group [80.95 ± 0.35% (ipsilateral); 79.82 ± 0.31% (contralateral)], and the NSCs-ic-24 h group [81.25 ± 0.45% (ipsilateral); 79.93 ± 0.36% (contralateral)], were not different from those of the ICH-vehicle group. These results indicate that intravenous transplantation at 24 h, or intracerebral. transplantation at 2 h or 24 h failed to reduce brain oedema. Only early intravenous transplantation effectively reduced the brain oedema.

Fig. 2

Brain water content and haemorrhage volume. Brain water contents were significantly lower in the ipsilateral (haemorrhagic) hemispheres of the NSCs-iv-2 h group than in all other groups at 3 days after ICH (A). Water contents of the other groups were not different from those of the ICH-vehicle group. NSC transplantations either intravenously or intracerebrally at 2 or 24 h did not influence haemorrhagic volume (B). Intravenous injection of rat neurosphere cells at 2 h (ICH-rat neurosphere-iv-2 h group) also reduced the mean brain water content in the ipsilateral hemisphere, while the human fibroblast injection (ICH-human fibroblast-iv-2 h group) showed no significant effect on the brain water content compared to the vehicle injection (C). **P < 0.01 versus all other groups.

According to the results of spectrophotometric assays of haemorrhage volumes at 3 days after ICH, NSC transplantation by either method (intravenous, intracerebral, 2 h and 24 h) exerted no influence on haemorrhage volumes (P > 0.05 in Kruskall–Wallis ANOVA; all P-values >0.05 versus ICH-vehicle, Fig. 2B), suggesting that equal burdens of haemorrhage were initiated.

To investigate whether allogenic neural progenitor cells can also reduce the brain oedema formation, we measured brain water contents in two more experimental groups made by intravenous injections of rat neurosphere cells (ICH-rat neurosphere-iv-2 h) or human fibroblasts (ICH-human fibroblast-iv-2 h) at 2 h after the ICH. The ICH-rat neurosphere-iv-2 h group showed the reduced brain water content in the ipsilateral hemisphere (80.05 ± 0.36%, P < 0.01 versus ICH-vehicle). However, the ipsilateral brain water contents of the ICH-human fibroblast-iv-2 h group (81.01 ± 0.36%) was not significantly different to that of the ICH-vehicle group (P = 0.078). In addition, the ICH-rat neurosphere-iv-2 h group, but not ICH-human fibroblast-iv-2 h group, showed the reduced initial neurologic deficits (MLPT score) compared to the ICH-vehicle group (ICH-rat neurosphere-iv-2 h: 4.00 ± 0.82, P < 0.01 versus ICH-vehicle; ICH-human fibroblast-iv-2 h: 6.2 ± 0.84, P = 0.104 versus ICH-vehicle). Accordingly, NSCs attenuated the brain oedema formation and initial neurologic deficits regardless of their allogenic or xenogenic transplantation conditions.

NSCs-iv-2 h reduced brain inflammatory infiltration and apoptotic cell numbers

Histologic quantifications of inflammatory infiltrations and apoptotic cells were undertaken in perihematomal areas to confirm the anti-inflammatory effects of NSCs-iv-2 h at 3 days after ICH (Fig. 3A). Initially, we observed intergroup differences among the five treatment groups in the inflammatory/apoptotic cell counts including OX42, MPO, Caspase-3 and TUNEL (P < 0.05, Kruskall–Wallis ANOVA), and further analyzed the values by intergroup comparison. It was found that the NSCs-iv-2 h group possessed significantly lower numbers of OX42+ microglia and macrophages than the ICH-vehicle group (P < 0.05), whereas the other groups showed no difference versus the ICH-vehicle group (ICH-vehicle: 105.0 ± 20.8 cells/mm2, NSCs-iv-2 h: 24.0 ± 6.32 cells/mm2, NSCs-iv-24 h: 79.8 ± 13.8 cells/mm2, NSCs-ic-2 h: 103.8 ± 14.4 cells/mm2, NSCs-ic-24 h: 113.5 ± 12.6 cells/mm2) (Fig. 3B).

Fig. 3

Histological analysis of inflammation and apoptosis. As representative photographs show (A), the NSC-iv-2 h group showed lower numbers of OX 42+ and MPO+ inflammatory cells and lower levels of activated caspase-3+ cells and TUNEL+ apoptotic cells in perihematomal regions at 3 days after ICH than the other four groups, as representative photographs show (B). The NSCs-iv-24 h group showed reductions of MPO+ and caspase-3+ cells versus the ICH-vehicle group. The # signs indicate hematoma areas. Bar = 100 μm. *P < 0.05 versus all other four groups. P < 0.01 versus the ICH-vehicle group.

MPO+ neutrophil counts in perihematomal areas revealed the most significant reduction in the NSCs-iv-2 h group (ICH-vehicle: 212.5 ± 29.9 cells/mm2, NSCs-iv-2 h: 44.3 ± 9.2 cells/mm2, NSCs-iv-24 h: 111.3 ± 19.7 cells/mm2, NSCs-ic-2 h: 190.3 ± 20.3 cells/mm2, NSCs-ic-24 h: 148.9 ± 66.6 cells/mm2, P < 0.05 between NSCs-iv-2 h and ICH-vehicle) (Fig. 3). The NSCs-iv-24 h group also showed a reduction in MPO+ count versus the ICH-vehicle group (P < 0.05), but this was less pronounced than of the NSCs-iv-2 h group (P < 0.05 versus NSCs-iv-24 h). However, the NSCs-ic-2 h and NSCs-ic-24 h groups showed no reduction in MPO+ neutrophil infiltration.

An analysis of active caspase-3+ apoptotic cell numbers showed that the NSCs-iv-2 h and NSCs-iv-24 h groups had lower counts than the ICH-vehicle group (all P < 0.05), and the NSCs-iv-2 h group had lowest counts than NSCs-iv-24 h (P < 0.05) (ICH-vehicle: 72.0 ± 10.6 cells/mm2, NSCs-iv-2 h: 17.8 ± 3.1 cells/mm2, NSCs-iv-24 h: 38.0 ± 6.7 cells/mm2, NSCs-ic-2 h: 76.5 ± 9.4 cells/mm2, NSCs-ic-24 h: 63.5 ± 11.0 cells/mm2) (Fig. 3).

TUNEL staining revealed a high density of positively-stained cells within and at the peripheries of haemorrhagic lesions (Fig. 3A). By quantitative analysis, the NSCs-iv-2 h group exhibited a significantly lower number of TUNEL+ cells (82.5 ± 43.7 cells/mm2) than the ICH-vehicle group (314.5 ± 50.8 cells/mm2; P < 0.05, Fig. 3B). However, the NSCs-iv-24 h (254 ± 44.6 cells/mm2), NSCs-ic-2 h (349.5 ± 63.6 cells/mm2), and NSCs-ic-24 h (325.5 ± 32.7 cells/mm2) groups were similar to the ICH-vehicle group. The perihematomal TUNEL+ cells were mostly positive for the neuronal marker (NeuN) (Supplementary Fig. 1A). Nevertheless, overall immunoreactivity against NeuN was diminished in the perihematomal area because of the loss of neuronal nuclear and/or cytoskeletal structure (Matsushita et al., 2000), and it seemed to be inaccurate to quantify TUNEL+NeuN+ cells.

Thus, we used Fluoro-Jade C staining to detect degenerating neurons in the perihematomal area (Wang and Tsirka, 2005). The ICH-NSCs-iv-2 h group had a lower number of Fluoro-Jade C stained neurons than the ICH-vehicle group at 3 days after ICH (ICH-vehicle: 176 ± 33 cells/mm2, ICH-NSCs-iv-2 h: 49 ± 18 cells/mm2, P < 0.05, Supplementary Fig. 1B). To confirm the sustained tissue protection, we also measured the hemisphere atrophies of the ICH-vehicle and the NSCs-iv-2 h groups at 35 days after ICH. The NSCs-iv-2 h group exhibited lesser degree of hemispheric atrophies than did the ICH-vehicle group (ICH-vehicle: 24.5 ± 3.5%, NSCs-iv-2 h: 16.0 ± 8.2%, P < 0.05). Accordingly, the NSCs-iv-2 h group demonstrated the most prominent and reproducible anti-inflammatory and neuroprotective effects.

NSCs-iv-2 h showed attenuated levels of cerebral and splenic inflammatory mediators

We then analysed changes in the levels of inflammatory mediators induced by NSCs-iv-2 h in brains, spleens and lymph nodes at 1 day after ICH. We have previously shown that ICH increases the expressions of TNF-α, IL-6, Fas and FasL in the brain samples at 1 day after ICH (Sinn et al., 2007). In the present study, we could confirm the upregulations of TNF-α, IL-1β, IL-4 and IL-6 in the haemorrhagic brains and spleens compared to the normal samples, with using RT-PCR supported by optical density analysis (all P < 0.05, n = 4 per group) (Fig. 4A and B). In lymph nodes, only IL-4 was upregulated (P < 0.05) and TNF-α, IL-1β, and IL-6 were unchanged after ICH. The levels of TGF-β were not affected by the ICH induction in all the three kinds of tissue.

Fig. 4

Changes in the expressions of cerebral and splenic inflammatory mediators. RT-PCR (A) and the optical density measure (B) revealed the upregulations of TNF-α, IL-1β, IL-4 and IL-6 in the haemorrhagic brains and spleens compared to the normal samples. In lymph nodes, only IL-4 was upregulated. The NSCs-iv-2 h group showed decreases of TNF-α, IL-4 and IL-6 mRNA levels in brain samples, as well as decreases of TNF-α and IL-6 mRNA levels in spleen samples (B). In the lymph nodes, the NSCs-iv-2 h injection attenuated the IL-4 upregulation. Western blotting for NF-κB expressions in the brain and spleen samples showed that the levels were significantly attenuated in the NSCs-iv-2 h group compared to the ICH-vehicle group (C). In the flow cytometric analysis of splenocytes, 24.2% of CD11b+ macrophages expressed TNF-α in the ICH-vehicle group, while 17.6% of CD11b+ macrophages expressed TNF-α in the normal spleen (D, from each single representative sample). In the NSCs-iv-2 h group, however, only 19.1% of CD11b+ macrophages expressed TNF-α. *P < 0.05 versus normal. P < 0.05 versus ICH-vehicle.

In brain samples, the NSCs-iv-2 h group showed decreases of 79% in TNF-α, 62% in IL-4, 47% in IL-6 mRNA levels versus the ICH-vehicle group (all P < 0.05, Fig. 4A and B). The NSCs-iv-2 h group also showed decreases in Fas and FasL mRNA levels compared to the ICH-vehicle group (by 35% and 66%, respectively, P < 0.05, Supplementary fig. 2A). In spleens, the NSCs-iv-2 h group showed decreases of 85% in TNF-α and 58% in IL-6 mRNA levels versus the ICH-vehicle group (all P < 0.05, Fig. 4A and B). In the lymph nodes, the NSCs-iv-2 h injection attenuated IL-4 levels by 69% (P < 0.05, Supplementary Fig. 2B). In addition, western blotting for NF-κB expressions in the brain and spleen samples showed that levels were significantly attenuated in the NSCs-iv-2 h group compared to the ICH-vehicle group (36% decrease in the brains and 81% in the spleens, all P < 0.05 versus ICH-vehicle, Fig. 4C).

To investigate whether NSCs-iv-2 h changed TNF-α expressions in the spleen macrophages, we performed flow cytometric analysis of splenocytes using anti-CD11b and anti-TNF-α antibodies. In the ICH-vehicle spleen, 24.2% of CD11b+ macrophages expressed TNF-α, while 17.6% of CD11b+ macrophages expressed TNF-α in the normal spleen (Fig. 4D). In the NSCs-iv-2 h group, however, only 19.1% of CD11b+ macrophages expressed TNF-α. In addition, the spleens from the NSCs-iv-2 h group showed reduced numbers of ED1+ activated macrophages in the marginal zone compared to those from the ICH-vehicle group (ICH-vehicle: 63 ± 9 cells/HPF, NSCs-iv-2 h: 30 ± 8 cells/HPF, p < 0.05, Supplementary Fig. 2C). In summary, these findings demonstrated that NSCs-iv-2 h reduced the levels of the important inflammatory mediators in both the brain and the spleen.

Distributions of injected NSCs

We previously observed that a number of NSCs migrated to perihematomal areas and differentiated into NeuN+, Neurofilament+, or GFAP+ cells in NSCs-iv-24 h group when we analyzed brains at 6 weeks after the transplantation (Jeong et al., 2003). However, at 1 day after the transplantation, only occasional NSCs were observed in a cerebral ischaemia model (Jin et al., 2005), although robust in situ proliferation was observed from 7 days after transplantation (Chu et al., 2004b). In the present study, we analyzed the distribution of injected NSCs in the NSCs-iv-2 h group at 3 days after ICH when their anti-inflammatory effects were apparent. However, only small numbers were observed in brain [≤1 cell per high-power field (HPF)], lung (2–3 cells per HPF), and liver (about 1 cell per HPF) (Fig. 5). This distribution is in accord with previous observations in an EAE model, where more NSCs were found to be distributed to the systemic organs rather than to the CNS just after transplantation (Pluchino et al., 2005). However, we observe few NSCs in mesenteric lymph nodes. In an EAE model, many transplanted NSCs were found in lymph nodes during the early stage, although they disappeared over several days (Einstein et al., 2007). However, in the present study, large numbers of NSCs were detected in spleen, especially in the marginal zone area (20–30 cells per HPF). Because this splenic zone contains many macrophages (Mebius and Kraal, 2005), we immunostained the spleen macrophages with anti-CD11b Ab. The results obtained confirmed that NSCs were located in the vicinity of marginal zone macrophages (Fig. 5F). In addition, we found that a number of NSCs were in cell-to-cell contact with CD11b+ spleen macrophages under the confocal microscopic views (Fig. 5G).

Fig. 5

Distributions of NSCs in vivo. NSCs were pre-labelled with DiO fluorescent dye before transplantation. In the NSCs-iv-2 h group, only a few NSCs were detected in brain (A), lung (B) and liver (C) at 3 days after the transplantation (arrows). No NSCs were observed in mesenteric lymph nodes (D). However, a number of NSCs are detected in spleen, especially in the marginal zone (MZ) (E). Immunostaining for CD11b demonstrated that NSCs in the spleen are in the vicinity of the marginal zone macrophages (F). We could observe that a number of NSCs were in cell-to-cell contact with CD11b+ spleen macrophages under the confocal microscopy (G, arrow heads, 1 μm step-size optical sections along the z-axis). WP = white pulp. Dotted line indicates the WP border. Bar = 100 μm in A–F, and 10 μm in G.

At 35 days after the transplantation, those NSCs-iv-2 h cells were still detected in lungs, livers and spleens (Supplementary Fig. 3A–E). However, we could observe few NSCs in the brains. The DiO-labelled cells were positive for human nuclear antigen, confirming that DiO -labelled cells were transplanted human cells (Supplementary Fig. 3F). About 80% of the NSCs observed in the spleen were positive for the cholinergic neuronal marker (ChAT) (Supplementary Fig. 3G and H). When we analyzed the distribution of NSCs in the NSCs-ic-2 h and NSCs-ic-24 h groups at 3 days after ICH, we observed abundant NSCs in the perihematomal area as well as transplantation foci of the brain, but not in the systemic organs (Supplementary Fig. 3I).

Splenectomy eliminated the effect of NSCs-iv-2 h transplantation

Because of the splenic bias shown by NSCs and the observed attenuations of inflammatory mediators in spleen, we hypothesized that this organ may facilitate inflammatory response modulation in our stroke model. To prove this, we performed splenectomy 3 days prior to the ICH induction, and then measured brain water contents at 3 days after ICH (n = 6 per each group). Mean brain water content was found to be 81.63 ± 0.55% in the ICH-vehicle group, and 80.10 ± 0.27% in the NSCs-iv-2 h group (P < 0.01, Fig. 6). Interestingly, the mean brain water content of splenectomized ICH rats (splenectomy-ICH-vehicle) was 80.8 ± 0.26% and was much lower than that of the ICH-vehicle group (P < 0.01), which suggests that the spleen is involved in cerebral oedema formation after ICH. Moreover, NSCs-iv-2 h transplantation in splenectomized ICH rats (splenectomy-ICH-NSCs-iv-2 h) resulted in a mean brain water content of 80.82 ± 0.25%, which was similar to that of the splenectomized ICH rats (splenectomy-ICH-vehicle) (P = 0.986). The mean brain water content of non-ICH normal control rats was 78.86 ± 0.30, and that of splenectomized non-ICH rats was 78.91 ± 0.13 (P = 0.714; no difference).

Fig. 6

Splenectomy eliminated the effects of early systemic NSCs transplantation. Splenectomy before ICH (Third bar, splenectomy-ICH-vehicle) reduced ipsilateral hemisphere water contents as compared with the ICH-vehicle group (First bar) (A). NSCs-iv-2 h transplantation in splenectomized ICH rats (Fourth bar, splenectomy-ICH-NSCs-iv-2 h) had similar levels of brain water content as splenectomy-ICH-vehicle rats (Third bar), which suggested that splenectomy eliminated the effects of NSCs. In histologic counts of perihematomal inflammatory infiltrations (B), the splenectomy-ICH-vehicle group showed less OX-42+ or MPO+ cells compared to the ICH-vehicle group, and NSCs-iv-2 h transplantation in the splenectomy-ICH rats (splenectomy-ICH-NSCs-iv-2 h) failed to further reduce the inflammatory cell numbers compared to the splenectomy-ICH-vehicle group. *P < 0.05, **P < 0.01. ns = not significant.

In histologic quantifications of perihematomal inflammatory cells, the splenectomy-ICH-vehicle group showed less OX-42+ or MPO+ cells compared to the ICH-vehicle group (all P < 0.05, Fig. 6B). However, NSCs-iv-2 h transplantation in the splenectomy-ICH rats (splenectomy-ICH-NSCs-iv-2 h) failed to further reduce the OX-42+ or MPO+ inflammatory cell numbers (all P > 0.05). In addition, there was also no difference in the initial neurologic scores (MLPT) at 1 day between the two groups (splenectomy-ICH-vehicle: 3.62 ± 0.52, splenectomy-ICH-NSCs-iv-2 h: 3.33 ± 0.52, P = 0.269), while the splenectomy-ICH-vehicle group showed less deficits compared to the ICH-vehicle group (P < 0.01). In summary, NSCs-iv-2 h transplantation failed to reduce brain oedema, perihematomal inflammatory cells and initial neurologic deficits in splenectomized rats, which indicated that the spleen plays a central role in the attenuation of ICH-associated brain injuries by intravenously administered NSCs.

NSCs inhibited in vitro macrophage activations via contact-dependency

TNF-α levels are one of the most important predictors of cerebral inflammation and oedema formation after ICH (Castillo et al., 2002). Because transplanted NSCs were found to localize in macrophage-rich splenic areas, we further focused on the interaction between NSCs and macrophages. Thioglycollate-elicited macrophages stimulated with lipopolysaccharide (LPS) (namely macrophage + LPS) robustly secreted TNF-α in vitro (referred to as 100 ± 1.5%, at 4 h after the LPS stimulation, Fig. 7). However, when macrophages are co-cultured with NSCs before LPS stimulation, TNF-α secretion induced by LPS stimulation significantly reduced to 70.9 ± 6.9% (P < 0.01 versus macrophage + LPS, n = 6 wells per group). However, when NSCs were cultured in transwell membrane inserts that prevented their contacting macrophages and allowed the diffusion of soluble factors, mean TNF-α secretion was 97.9 ± 1.8%, i.e., no different to the macrophage + LPS situation (P = 0.149). These results suggest that NSCs can inhibit in vitro macrophage activations via a contact-dependent mechanism.

Fig. 7

NSCs inhibited in vitro macrophage activation via cell-to-cell contact. Macrophages stimulated with LPS robustly secreted TNF-α (second bar), and macrophage/NSC mixed-cultures showed lower TNF-α levels in culture medium (third bar) at 4 h after the LPS stimulation (A). However, this effect was eliminated when NSCs were cultured in transwell membrane inserts which prevented cell-to-cell contact and permitted only the diffusion of soluble factors (fourth bar). NSCs alone showed no response to LPS stimulation (fifth bar). When we added NSCs to the cultured macrophages at 2 h after the LPS stimulation and measured cytokine levels at 24 h after the LPS stimulation, NSCs still attenuated the TNF-α secretion of macrophages (B). IL-6 secretion were not inhibited by NSCs. *P < 0.05, **P < 0.01. ns = not significant.

As an adjuvant experiment to simulate the in vivo condition of NSCs-iv-2 h transplantation after ICH, we added NSCs to the cultured macrophages at 2 h after the LPS stimulation and measured cytokine levels at 24 h after the LPS stimulation. In this experiment, NSCs again attenuated the TNF-α secretion of macrophages by 56% (P < 0.05, n = 3 wells per group, Fig. 7B). However, the levels of IL-6 secretions were not changed by NSCs co-culture (P = 0.369).


In this study, we undertook to characterize the anti-inflammatory effects of NSCs, and the mechanisms involved. Intravenous NSC transplantation during the hyperacute stage (at 2 h after ICH) was found to reduce initial neurologic deterioration, and to exhibit anti-inflammatory and anti-apoptotic properties with attenuating brain oedema and the expressions of inflammatory mediators in brain and spleen. Unexpectedly, we also observed that the spleen participates in cerebral inflammation as splenectomy reduced cerebral oedema and inflammatory cell counts, and that NSCs modulate the splenic inflammatory pathway to reduce the cerebral inflammation. In addition, NSCs inhibited in vitro macrophage activation via a contact-derived mechanism.

Our results suggest that earlier intravenous NSC administration can attenuate systemic inflammatory response after haemorrhagic stroke and protect the brain via a bystander mechanism rather than via any direct cell replacement. To date, NSCs have been transplanted to regenerate damaged brain tissue, and previously, we showed that human NSCs administered intravenously at 24 h after ICH promoted neurological recovery and replaced lost neuronal cells (Jeong et al., 2003). Moreover, suppressions of T-lymphocyte activity by mesenchymal stem cells or neural precursors have been demonstrated in EAE models (Pluchino et al., 2003, 2005; Zappia et al., 2005; Einstein et al., 2007; Gerdoni et al., 2007), and it has been suggested that the immunomodulatory properties are a characteristics of stem cells rather than neural cells (Miller and Bai, 2007). However, the present study is the first to report that intravenous NSCs administered during the hyperacute stage in stroke can modulate innate cerebral inflammatory responses with interacting with peripheral inflammatory systems.

As demonstrated by this study, the spleen is important both as an inducer of cerebral inflammation and as a target for NSC-based intervention. The link between brain and spleen inflammation can be referred to as ‘brain-spleen inflammatory coupling’. Humoral pathways facilitate brain/systemic immunity communication. Cytokines like TNF-α, IL-1β and IL-6 secreted by cells in different tissues and organs link their inflammatory responses (Chamorro et al., 2007). In addition, neural and hormonal mechanisms like the cholinergic anti-inflammatory pathway, the adrenergic pathway and the hypothalamic pituitary axis (HPA) link the brain and peripheral immune organs (Prass et al., 2003; Meisel et al., 2005; Tracey, 2007). Thus, changes in the peripheral immune system can occur within hours of brain injury, and may later cause stroke-induced immune depression syndrome (SIDS) or increase the risk of infection (Dirnagl et al., 2007). In the present study, reductions in splenic inflammation by NSCs without significant migration to the brain were coupled with a reduction in brain inflammation and a favourable neurologic outcome. In addition, splenectomy prior to ICH reduced the cerebral oedema formation and the number of perihematomal inflammatory cells per se. Accordingly, modulations of spleen-mediated inflammation may provide new solutions for cerebral anti-inflammatory treatment and brain protection.

The splenic marginal zone is an important transit area for cells that are leaving the bloodstream and entering the white pulp. Moreover, based on spleen structure, most blood flows through the marginal zone and directly along the white pulp, which leads to efficient monitoring of the blood by the peripheral immune system (Mebius and Kraal, 2005). Macrophages represent a major component of the marginal zone and express pattern-recognition receptors (e.g. toll-like receptors) that are utilized for pathogen and antigen clearance (Mebius and Kraal, 2005). Accordingly, early splenic inflammatory activation after stroke (Offner et al., 2006) and the localization of systemically injected NSCs in the splenic marginal zone well correspond.

TNF-α activation can be detected a few hours after stroke; even before significant neuronal death has occurred (Allan and Rothwell, 2001), and the spleen secretes robust amounts of TNF-α into circulating blood at this early stage (Offner et al., 2006). TNF-α promotes early-stage inflammation by increasing the expressions of chemotactic factors and adhesion molecules by vascular endothelium, which leads to the early infiltration of monocytes/macrophages, neutrophils, and T cells into sites of injury (Barone and Feuerstein, 1999; Stevens et al., 2002). Early modulation of inflammatory cell response by administering an antibody blocking TNF-α at 2 h after stroke reduced brain damage and improved neurologic outcomes in an experimental model (Hosomi et al., 2005). Accordingly, interventions targeting splenic TNF-α activation require a prompt approach. Our result that NSC transplantation at 24 h after ICH had no attenuating effect on brain oedema/inflammation confirm this idea, because TNF-α has probably been activated systemically and centrally by this time, as shown by the present study and others (Offner et al., 2006; Sinn et al., 2007). Thus, the observed differences between NSC-iv-2 h and NSC-iv-24 h probably depend on the pathogenic time course.

The homing of NSCs to damaged brain is dependent on stromal cell derived factor-1α (SDF-1α)-CXC chemokine receptor 4 signalling, or on adhesion molecule-receptor signalling (Imitola et al., 2004; Mueller et al., 2006). Although the expressions of SDF-1α and other adhesion molecules in brain might be significant at later times, brain homing signals might not be significant at 2 h after ICH when NSCs-iv-2 h cells were injected in the present study. In addition, blood-brain barrier might be less penetrable for NSCs at this time window. This might answer why NSCs-iv-2 h cells failed to migrate to the brain. However, migration of NSCs to spleen might not be dependent on inducible factors. The H1 cell line used in this study highly expresses CD44+ (Chu et al., 2004b), which is also expressed on inflammatory cells and mediates their infiltration via CD44/hyaluronan interaction (Pure and Cuff, 2001). In an EAE model, intravenously administered neurosphere cells were predominantly detected in lymph nodes at 1 day after transplantation, and were not detected in brain, spinal cord, or even in spleen (Einstein et al., 2007). These two different models, i.e. the haemorrhagic stroke and the EAE models, with predominantly innate and adaptive immune responses, respectively, may affect the biodistribution patterns and immunologic effects of neural stem cells. Although the NSCs-iv-2 h treatment also reduced IL-4 levels in ICH brains and lymph nodes in the present study, IL-4 is an anti-inflammatory molecule, mainly secreted by a negative feedback mechanism responsive to the other inflammatory cytokines (Tedgui and Mallat, 2001), and is not associated with the clinical outcome of stroke (Vila et al., 2003). Accordingly, further studies are warranted to reveal how NSCs migrate to immune organs and how they interact with inflammatory cells, although the present study demonstrates initial evidences that the mechanism of macrophage inactivation by NSCs involves a contact-dependent interaction.

The systemic transplantation of NSCs is probably the least invasive method of cell administration. Intracerebral direct neural transplantation requires invasive surgery and leaves needle tracks and mass effects. In human, the local stereotaxic infusion of NSCs may be severely limited because infused cells must migrate substantial distances and the time taken to embrace entire disease sites may be unacceptable. Thus, to overcome these obstacles, NSCs must be administered at numerous sites. In contrast, an intravenous injection can disperse NSCs according to natural cues and allow them to migrate along the host's ‘chemotactic gradients’ (Pluchino et al., 2003; Chu et al., 2004b; Martino and Pluchino, 2006). A similar systemic approach has been studied using hematopoietic stem cells in stroke models (Taguchi et al., 2004; Nomura et al., 2005; Honma et al., 2006). Moreover, the transplantation of NSCs into circulating blood may be easier and more effective in multifocal and large-lesioned CNS disorders (Jeong et al., 2003; Pluchino et al., 2003; Chu et al., 2004d; Fujiwara et al., 2004; Pluchino et al., 2005). In addition, based on the present study, the migration of NSCs to the spleen can be therapeutically beneficial and be another major contribution in stroke models. Nevertheless, the immune responses in the brain and other systemic organs are complex, and cannot be ignored when it comes to repair strategies involving cellular transplants (Barker and Widner, 2004). Accordingly, further studies are warranted in this issue.

Collectively, our results suggest that systemic injections of human NSCs during hyper-acute stage ICH have potential therapeutic relevance, because they were found to display strong anti-inflammatory functions that promote secondary neuroprotection by interrupting splenic inflammatory response. The present study demonstrates that the spleen is centrally involved in innate immune response after ICH, and thus, represents an important target for the therapeutic development of stem cell therapies in acute stroke.

Supplementary material

Supplementary material is available at Brain online.


This research was supported by a grant (SC3060) from Stem Cell Research Center of the 21st Century Frontier Research Program funded by the Ministry of Science and Technology, Republic of Korea.


  • *These authors contributed equally to this work.

  • Abbreviations:
    basic fibroblast growth factor
    choline acetyltransferase
    Dulbecco's Modified Eagle's Medium
    experimental autoimmune encephalitis
    epidermal growth factor
    glial fibrillary acidic protein
    high power field
    intracerebral haemorrhage
    modified limb placing test
    nuclear factor-kappaB
    neural stem cells
    stromal cell derived factor-1alpha
    tumour growth factor
    tumour necrosis factor-alpha
    terminal deoxynucleotidyl transferase biotin-dUTP nick end labelling


View Abstract