Abstract
Experimental transplantation of human umbilical cord blood (hUCB) mononuclear cells (MNCs) in rodent stroke models revealed the therapeutic potential of these cells. However, effective cells within the heterogeneous MNC population and their modes of action are still under discussion. MNCs and MNC fractions enriched (CD34+) or depleted (CD34-) for CD34-expressing stem/progenitor cells were isolated from hUCB. Cells were transplanted intravenously following middle cerebral artery occlusion in spontaneously hypertensive rats and directly or indirectly cocultivated with hippocampal slices previously subjected to oxygen and glucose deprivation. Application of saline solution or a human T-cell line served as controls. In vivo, MNCs, CD34+ and CD34- cells reduced neurofunctional deficits and diminished lesion volume as determined by magnetic resonance imaging. MNCs were superior to other fractions. However, human cells could not be identified in brain tissue 29 days after stroke induction. Following direct application on postischemic hippocampal slices, MNCs reduced neural damage throughout a 3-day observation period. CD34+ cells provided transient protection for 2 days. The CD34- fraction, in contrast to in vivo results, failed to reduce neural damage. Direct cocultivation of MNCs was superior to indirect cocultivation of equal cell numbers. Indirect application of up to 10-fold MNC concentrations enhanced neuroprotection to a level comparable to direct cocultivation. After direct application, MNCs migrated into the slices. Flow cytometric analysis of migrated cells revealed that the CD34+ cells within MNCs were preferably attracted by damaged hippocampal tissue. Our study suggests that MNCs provide the most prominent neuroprotective effect, with CD34+ cells seeming to be particularly involved in the protective action of MNCs. CD34+ cells preferentially home to neural tissue in vitro, but are not superior concerning the overall effect, implying that there is another, still undiscovered, protective cell population. Furthermore, MNCs did not survive in the ischemic brain for longer periods without immunosuppression.
Keywords
Introduction
Transplantation of human umbilical cord blood (hUCB) mononuclear cells (MNCs) for experimental stroke therapy has been thoroughly investigated since a therapeutic impact of these cells was first reported (8). Numerous hUCB cell-mediated effects on postischemic cerebral tissue have been discovered, but the precise therapeutic mechanisms of cord blood cells applied after stroke still remain unclear. Intravenous administration reduced infarct volume and ameliorated behavioral deficits following middle cerebral artery occlusion in rodents (40). With special relevance for stroke therapy, hUCB MNCs are effective when given in a clinically relevant time frame of 48 h (20), which is far beyond the 4.5-h time window for thrombolysis (13), currently being the only effective stroke therapy in humans. Cord blood cell populations are further expected not to induce severe graft versus host disease in allogenic use. Cord blood cells are also among the few stem cell containing populations allowing for autologous application and might therefore become of clinical relevance (24,28,45).
The MNC fraction of hUCB is a heterogeneous population that was previously characterized elsewhere (35). Characteristically, hUCB MNCs contain myeloid cells (CD14+, about 14%) and lymphocytes (86%). Most relevant lymphocyte subpopulations are B-cells (CD19+, about 12%), T cells (CD3+, about 73%), and natural killer cells (CD56+, about 13%). The remaining 2% of cells within hUCB MNCs are considered stem and progenitor cells with a majority being CD34+/CD133+ or CD34+/CD133-. Following experimental stroke systemically delivered MNCs were identified near the ischemic lesion by immunohistochemistry and polymerase chain reaction (40). However, Chen et al. (8) localized only 0.3% of administered MNCs near the ischemic lesion with rare signs of differentiation into neuronal tissue. A final proof for in vivo differentiation of hUCB cells into functional neurons following stroke is also lacking. Thus, it is more likely that cells influence the imbalanced neuronal microenvironment after stroke than differentiating into functional brain tissue (21). There is growing evidence that UCB cell therapy may exert neuroprotection by anti-inflammatory and immunomodulating effects (18,19). MNCs are also known to secrete numerous soluble factors as interleukin (IL)-8, macrophage chemoattractive protein-1 (21), IL-10 (26), and nerve growth factor (NGF) (6). Supplying these factors, therapeutic cells may protect poststroke tissue by limitation of free radicals and inflammatory cytokines as well as by neurotrophic signaling (21,29). Umbilical cord blood is furthermore a source of endothelial progenitor cells (39), possibly enhancing revascularization in subacute phases after ischemic stroke.
Interestingly, presence of MNCs in the lesioned brain is not even a prerequisite for neuroprotection (5), raising the question of additional therapeutic benefits by close spatial interaction of transplanted cells and ischemic tissue. Despite encouraging results of hUCB administration following stroke in preclinical trials, more knowledge about the effective cell population within hUCB MNCs is needed before moving on to clinical studies.
In this study, we comparatively examined hUCB MNC, CD34+, CD34- fractions, and human Jurkat T cells (hT cells) regarding their ability to exert neuroprotection both in vivo and in vitro. Reduction of functional deficits and lesion size following middle cerebral artery occlusion (MCAO) was monitored in spontaneously hypertensive rats (SHR). These animals suffer from various stroke-relevant comorbidities and are recommended for preclinical stroke research (38). Additionally, rodent hippocampal slice cultures were used to determine effects of particular cell populations on neuronal survival after oxygen and glucose deprivation (OGD), also considering cell migration, direct and indirect application paradigms, and dose escalation effects.
Materials and Methods
Preparation of Umbilical Cord Blood Samples
Informed consent cord blood sample donations of healthy full-term neonates were collected immediately after delivery. Sample donations were approved by the University of Leipzig Regional Ethical Board. MNCs were obtained by density gradient separation and cryopreserved in fetal calf serum containing 8% dimethyl sulfoxide (DMSO; Serumwerke Bernburg, Germany). Prior to use, MNCs were rapidly thawed in phosphate-buffered saline (PBS) containing 75 U/ml DNase I (Roche Diagnostics, Germany) and rinsed in RPMI-1640 (PAA, Germany).
CD34+ cells were isolated from the MNC fraction by using the MACS® immunomagnetic positive selection protocol (Miltenyi Biotech, Germany). Obtained cells were analyzed using a FACSCalibur flow cytometer equipped with the CellQuest™ software (both Becton-Dickinson, USA) and characterized by anti-CD45-fluorescein isothiocyanate (FITC) and anti-CD34-allophycocyanin antibodies (APC, both Beckman Coulter, Germany). Isotype-identical monoclonal antibodies served as controls. After separation, CD34-enriched (CD34+) and CD34-depleted (CD34) fractions contained 94.11 ± 3.42% or 0.06 ± 0.03% CD34-expressing cells, respectively.
MNC, CD34+, and CD34- cell suspensions were labeled with carboxy fluoresceindiacetate succinimidyl ester (CFSE, 5 μM; Molecular Probes, The Netherlands) for 10 min at 37°C prior to transplantation. For in vitro migration studies, MNCs were stained with CellTracker Orange (CTO, 5 μM; Molecular Probes) according to the manufacturer's protocol.
Human Jurkat T Cells
Cryopreserved human clonal Jurkat T cells [hT cells, stored in RPMI-1640/10% fetal calf serum (FCS) supplemented with 10% DMSO] were donated by the Faculty of Biology, Pharmacy and Psychology, University of Leipzig. Immediately before use, cells were thawed and viability was assessed by trypan blue exclusion test. Cell staining by CFSE or CTO was performed as described above.
Middle Cerebral Artery Occlusion (MCAO) and hUCB Cell Transplantation
All experiments involving animals were approved by the local governmental commission for animal welfare. Forty adult male spontaneously hypertensive/NCrl rats (SHR, Charles River, Germany) were randomly assigned to five experimental groups (n = 8 each) and subjected to permanent cortical MCAO. Briefly, animals were anaesthetized by intraperitoneal injection of ketamine (100 mg/kg, Merial, Germany), xylacine (10 mg/kg, Bayer, Germany), and atropine (0.1 mg/kg, Ratiopharm, Germany). Throughout the surgical procedure, core body temperature was continuously measured by a rectal probe and kept constant at 37°C using a feedback controlled warming pad (Fine Science Tools, Germany). O2-enriched (35%) air was applied for breathing. After incision of the temporal muscle, the left MCA was exposed by trepanation and local dura incision, and subsequently cauterized. After wound sealing, animals were transferred back to their home cages and allowed to recover. Twenty-four hours after MCAO, animals received intravenous injection of 1.0 ml PBS containing either 1 × 106 hUCB MNCs, CD34+, CD34-, hT cells, or no cells (vehicle control) into a tail vein.
Behavioral Assessment
After training for 7 consecutive days, RotaRod motor functional tests were performed by an investigator blinded to the experimental groups. Mean baseline values were recorded for each animal from five individual trials conducted 24 h prior to MCAO. Mean was set 100%. Subjects were placed on the cylinder that initially rotated with four rounds per minute (rpm). Rotating speed was then constantly accelerated to a maximum of 40 rpm within 3 min. A trial ended if an animal fell off the rung, gripped it and spun around for three revolutions without walking, or if the rat kept running for 5 min. Five RotaRod tests were performed at days 1, 2, 4, 7, 11, 17, 23, and 29 after stroke. Mean values were calculated in relation to baseline performance.
Magnetic Resonance Imaging (MRI)
Five individuals from each experimental group were randomly chosen to undergo MRI investigation at 20 h (before cell transplantation), and 7 and 29 days following MCAO. Animals were anesthetized as described above and placed on a laboratory-made rack that allowed positioning of the animal in the center of the static magnetic field. MRI was performed using a clinical standard Gyroscan Intera 1.5T scanner equipped with a 47-mm RF-coil (both Philips, The Netherlands). Spontaneous respiration was monitored throughout imaging by the breathing sensor of the scanner, which was placed under the subject's abdomen. T2-weighted turbo spin echo sequences were performed consisting of 20 transverse slices (1 mm thickness; matrix: 244 × 244; field of view: 50 μm). For analysis of ischemic damage, areas of the diffusion disturbance and both ipsilateral and contralateral ventricles were calculated slice-wise and interpolated to a volume using the Image J software and the YAWI 2D 1.3.2. selection tool (open source). Lesion size was estimated by the volume of the diffusion disturbance added to the differences between the ipsilateral and contralateral ventricle. Day 1 values were set 100%. As a parameter for stroke-induced tissue loss, hemispherical atrophy was calculated from T2 images on day 29. For this, the volume of the ipsilateral hemisphere minus infarct and associated side ventricle was divided by the volume of the contralateral hemisphere minus side ventricle.
Brain Tissue Analysis
After the last MRI on day 29 rats were sacrificed in deep anesthesia by transcardial perfusion using 40 ml of PBS followed by 80 ml of 4% paraformaldehyde (PFA). Brains were removed after decapitation and kept in 4% PFA overnight. Over the next 3 days brains were stored in 10%, 20%, and 30% sucrose (Sigma Aldrich, USA) for 24 h each, and thereafter stored at −80°C. Coronal slices (30 μm thick) were cut using a cryostat (Leica Microsystems, Germany) and mounted on Superfrost® slides (Resolab Inc., Germany). Screening of brain slices for transplanted, CFSE-positive cells was performed using an Axioplan fluorescence microscope (Carl Zeiss, Germany).
Organotypic Hippocampal Slice Cultures
Organotypic hippocampal slice cultures were prepared from 60 postnatal Wistar rat pups (P7–9, Harlan Winkelmann, Germany) as described previously (34) or, for migration experiments, from 15 neonatal transgenic B6. Cg-TgN (Thy1-YFP) 16 Jrs mice (P5–9, Charles River, Germany). The latter animals express high levels of enhanced yellow fluorescent protein (eYFP) in neuron subsets, including pyramidal cells of the hippocampus (11). Pubs were sacrificed by decapitation and the brain was removed for hippocampus preparation. Hippocampi were dissected and transversely sliced (350 μm) on a McIlwain tissue chopper (The Mickle Laboratory Engineering, UK). Slices were then carefully transferred onto humidified 0.4-μm porous Millicell membrane inserts (Millipore, France) and maintained in 1 ml of serum-based medium (50% MEM-Hanks, 25% HBSS, 17 mM HEPES, 5 mM glucose, 1 mM l-glutamine, 25% horse serum, and 0.5% gentamycin) at 37°C for 3 days. Thereafter, cultures were transferred to serum-free medium (horse serum replaced by 25% Neurobasal-A and 0.5% B27; referred to as standard condition) and kept for 7 days at 37°C in 5% CO2. Twenty-four hours before experimental use, 2 μl propidium iodide (PI; 1 mg/ml, Sigma, Germany) was added to the culture medium to exclude predamaged slices.
Oxygen and Glucose Deprivation Injury
For OGD, hippocampal slices were transferred to six-well plates containing 1 ml of 10 mM mannitol-containing glucose-free Ringer solution. Cultures were placed in a gas-tight chamber at 37°C. Oxygen was replaced by 95% N2 and 5% CO2 within 10 min and slices were incubated for another 40 min. Untreated control cultures were maintained under standard conditions (see above) in a normoxic atmosphere for the same time.
Coculturing of Hippocampal Slices with hUCB Cells
Slices were returned to normoxic standard culture conditions after OGD and 1 μl Neurobasal A medium containing 2.5 × 104 MNCs (CTO labeled for migration studies), CD34+, CD34- cells, or hT cells was administered directly onto the slices. MNCs were administered to the center of the slice using a Hamilton microliter syringe without targeting a specific region. Due to surface tension cell suspension spread immediately throughout the slice surface. For indirect coculturing, 2.5 × 104, 12.5 × 104, or 25 × 104 MNCs per slice were seeded under a membrane. Direct and indirect cocultures were analyzed over a period of 3 consecutive days.
Analysis of Cellular Damage
Damage was analyzed in the cornu ammonis (CA1-CA2-CA3) after slices have been incubated for 2 h with 1 mg/ml PI at 22, 46, or 70 h post-MNC administration. PI fluorescence was thereafter elicited at 546 nm and recorded at >610 nm on an inverted Nikon Eclipse TE-2000 fluorescence microscope (4x magnification; Nikon, Germany) and analyzed using an image analysis software (LUCIA M, Nikon). To ensure comparability of data fluorimetric mean values of untreated OGD controls obtained at day 1 were defined as 100%. All other data obtained were given as relative values.
Descriptive Migration Assay and Quantitative Flow Cytometric Analysis of Migrating MNCs
MNCs were stained with CTO as stated above. Migration studies were performed using OGD and normoxic control slices after 3 days of maintenance using a Zeiss LSM Pascal 5 confocal laser scanning microscope (Zeiss, Germany; Plan Neofluar 20x or 40x).
For quantitative migration assessment, slices were collected and stored at −80°C until further use. After thawing, slices were homogenized using 70-μm cell strainers (BD Falcon, USA). Extracts were dissolved in serum-free medium and protein concentration was adjusted to 1 μg/ml. Five hundred microliters of homogenates was filled in the lower compartment of the migration chambers. Equal volumes of medium, supplemented with 200 ng/ml recombinant stromal cell-derived factor (SDF; Immunotools, Germany) served as controls. MNCs (1 × 106) were suspended in 100 μl medium and administered into the upper part of the migration chamber on 8-μm pore size cell culture inserts (Greiner, Germany). This pore size requires active migration for the cells to reach the lower compartment. Cells were allowed to migrate for 4 h at 37°C and 5% CO2 before inserts were removed and migrated MNCs were collected. Viability and numbers of migrated cells were determined using the trypan blue exclusion method and BD Truecount Tubes (BD Biosciences), respectively. MNCs were characterized according to cellular subtypes by use of a FACSCalibur flow cytometer (Becton Dickinson) using the following sets of antibodies: (i) CD45-APC (leukocyte common antigen), the B-cell marker CD19-phycoerythrin-cyanin 7 (PE-Cy7, both Beckman Coulter), CD14-FITC (macrophage marker, BD), CD3-PE (T-cell receptor, Immunotech, Germany), (ii) CD45-PE-Cy7, CD133/2-PE, CD34-APC (stem and progenitor cell markers, Beckman Coulter), (iii) CD3-FITC, CD56-PE (natural killer cells, BD). Isotype-identical monoclonal antibodies served as controls. FACS experiments were performed in accordance to the International Society of Hematotherapy and Graft Engineering (ISHAGE) protocol (36). The trypan blue and Truecount Tube viability assays were performed immediately before the FACS experiments and so no additional live/dead gating was carried out.
Analysis of Human- and Rat-Soluble Mediators
A total volume of 1 ml of coculture media supernatant per well was analyzed using an antibody array kit (RayBiotech, USA) to profile human and rat trophic factors. Membranes for analysis of soluble factors were plotted with antibodies against human brain-derived neurotrophic factor (BDNF), NGF, epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), and insulin-like growth factor 1 (IGF-1), or against rat ciliary neurotrophic factor (CNTF), NGF, and VEGF. For methodological reasons, supernatants were pooled from six wells, each containing three slices. Samples were centrifuged (400 x g, 10 min) and stored at −80°C. Double concentrated (3,400 x g, 20 min, Amicon Ultra-4, Millipore, Germany) supernatants from OGD cultures as well as supernatants after direct (2.5 × 104) and indirect (2.5 × 104, 12.5 × 104, 25 × 104) application of MNCs were investigated. Array analysis was performed according to manufacturer's instructions. Chemiluminescence of membranes was visualized in an Intelligent Dark Box II (Fuji Photo Film, Germany) for 40 s.
Statistical Analysis
All results are reported as mean ± SD except for data on MNC migration and rat-soluble factors, which, for methodological reasons, represent investigation of pooled samples. Analysis was performed using the SPSS statistical software (version 11.5.1). After testing for Gaussian distribution, data of two independent variables were analyzed using Student's t-test for equally distributed data or the Mann-Whitney test (not normally distributed data), respectively. For three or more variables, ANOVA or ANOVA on ranks was performed followed by post hoc analysis using Tukey's test. Values of p < 0.05 were considered to be statistically significant and are indicated by single asterisks or alternative symbols (*, #, and §).
Results
Improvement of Motor Function After Transplantation of hUCB Cell Fractions
MCAO caused distinct motor deficits as indicated by a pronounced decrease of RotaRod performance on day 1. Control animals, but not hT cell-treated subjects, showed a slow and limited spontaneous recovery over time. Functional deficits were more effectively diminished in all subjects treated with hUCB-derived cell populations. This was evident from day 17 onwards for hUCB MNCs (p < 0.05 or less) (Fig. 1A) and from day 7 onwards for CD34+ and CD34- (p < 0.05 or less) (Fig. 1B, C), respectively, until the trial endpoint. Performance of hT cell-administered animals was inferior to all hUCB-treated subjects from day 7 on (p < 0.01), as well as compared to control subjects on day 11 (p < 0.05) and at the end of trial (p < 0.01) (Fig. 1A–D). Groups that received CD34+ and CD34- cell treatment performed very similarly throughout the observation period and almost reached baseline level on day 17 (Fig. 1B–D). hUCB MNC injected rats were even superior to CD34+- and CD34--treated subjects from day 17 onwards (p < 0.05 or less) (Fig. 1D).
Functional assessment using the RotaRod. (A–C) Administration of human umbilical cord blood (hUCB) cell fraction effectively reduced motor deficits in all groups (n = 8 each) compared to phosphate-buffered saline/human clonal Jurkat T (PBS/hT) cell application from day 7 [CD34+ (B); CD34- (C); p < 0.05 or less], or from day 17 [mononuclear cells (MNCs) (A); p < 0.05 or less]. After day 17, MNC administration was even superior to CD34+ and CD34- treatment [p < 0.05 or less (D)]. hT cell injection worsened motor functions in comparison to PBS treatment at days 11 (p < 0.05) and 29 (p < 0.01) (A–C). *p < 0.05 versus MCAO, #p < 0.05 versus hT cells, §p < 0.05 versus CD34+, +p < 0.05 versus CD34-; **p < 0.01 versus MCAO, ##p < 0.01 versus hT cells, §§p < 0.01 versus CD34+, ++p < 0.01 versus CD34-.
hUCB Administration Reduces Lesion Size and Hemispherical Atrophy After MCAO
No differences were detected in absolute lesion volumes on day 1 (data not shown). In all experimental groups, there was a relative reduction of lesion size on day 7 and day 29 post-MCAO (Fig. 2A) in relation to the lesion in T2 images obtained on day 1. No differences between groups were evident on day 7 (p > 0.4). Lesion volume reduction on day 29 was more evident in groups that received hUCB subfractions compared to PBS and hT cell treatment (p < 0.01) (Fig. 2A). There was no significant difference between groups that were administered hUCB subsets or between the control and the hT-cell group.
MRI-based investigation of lesion volume and hemispherical atrophy. Administration of hUCB cell populations resulted in smaller lesions at day 29 (p < 0.01), but not at day 7 (p > 0.4) compared to PBS and hT cell injection (A). Hemispherical atrophy at day 29 was lowest in MNC-treated subjects, being significantly different from CD34+, PBS, and hT cell treatment (p < 0.05 or less) (B), but not to CD34- application (p = 0.057) (B). Representative T2 slice series are given for day 1 (C) and day 29 (D, E). Note massive enlargement of ipsilateral ventricles after PBS and hT cell application in contrast to moderate extension in HUCB-treated groups. *p < 0.05 versus MCAO, #p < 0.05 versus hT cells, §p < 0.05 versus CD34+; **p < 0.01 versus MCAO, ##p < 0.01 versus hT cells. Scale bars: 5 mm (C–E).
Hemispherical atrophy, as an indicator of MCAO-caused cerebral tissue loss, was most prominent in hT cell-treated (15.4 ± 4.5%) and control (14.0 ± 3.6%) subjects compared to hUCB cell subfraction-treated animals (p < 0.05 or less) on day 29 (Fig. 2B). Administration of hUCB MNCs (5.2 ± 2.2%) diminished tissue loss more effectively than CD34+ cell application (9.6 ± 2.2%, p < 0.05), but level of significance was not reached in comparison to CD34- treatment (8.2 ± 2.3%, p = 0.057). Figure 2 also shows representative series of T2-weighted MR images at 20 h (Fig. 2C, prior to cell transplantation) and on day 29 of hUCB cell-injected animals (Fig. 2D) as well as PBS/hT cell-treated subjects (Fig. 2E). No signs of secondary bleeding, volume-consuming processes (like tumors), or infections were observed in any group at any time point.
Detection of CFSE-Positive Cells From the Transplant In Vivo
No CFSE fluorescence indicating transplanted cells could be detected in any subject of any group on day 29 (data not shown). Thus, cells were most likely not present in the recipient brains 1 month after transplantation.
Reduction of OGD-Induced Damage in Hippocampal Slices After Cell Application
With this set of experiments we evaluated the protective properties of different cell preparations derived from hUCB (MNC, CD34+, CD34) and of a hT cell line by semiquantitative analysis of PI staining after OGD. As indicated by a strong increase in PI staining, OGD induced broad cellular damage, which was found predominantly in the pyramidal layer of the hippocampal slices. The degree of damage in untreated OGD controls remained stable throughout the whole exploration period (p > 0.95) (Fig. 3A, B). Direct application of MNCs had a significant protective impact (p < 0.05) and reduced cellular degeneration at every investigated time point (Fig. 3A, B).
Evaluation of neuroprotective properties of MNCs, CD34+, CD34-, and hT cells following OGD in hippocampal slices. After oxygen-glucose deprivation (OGD), 2.5 × 104 cells of each fraction were directly applied onto OGD slices. Cellular degeneration in the cornu ammonis was assessed by propidium iodide (PI) fluorescence over 3 consecutive days in three separate experiments comprising six slices each. Fluorimetric values obtained from OGD slices without treatment at day 1 were set 100%. (A) All groups were compared to OGD without treatment (black bar) at day 1. Cellular damage remained stable in nontreated OGD controls (black bars). Application of MNCs and CD34+ cells diminished damage in comparison to OGD control at days 1 and 2 (p < 0.05) as indicated by reduced PI fluorescence intensity, being even indifferent from normoxic controls at these days (white bars, p > 0.4). In contrast, only MNCs were able to provide protection until the end of the observation period. CD34- cells failed to prevent cell damage in the cornu ammonis while the application of hT cells even aggravated cellular damage at day 3 (p < 0.05). White bars represent PI fluorescence in normoxic control hippocampal slices. All bars represent relative mean values ± SD. (B) Representative fluorescence micrographs of rat hippocampi cultivated under normoxic conditions, after OGD with and without by MNC application. *p < 0.05 versus OGD without treatment (black bar). Scale bar: 500 μm.
Comparable to MNCs, the CD34+ fraction reduced cell degeneration on day 1 and day 2 but failed to provide protection on day 3. On the other hand, depletion of CD34-expressing cells (CD34-) from the MNC fraction abolished reduction of neural damage, as indicated by more intense PI staining. When applied on damaged hippocampal slices hT cells also did not reduce neural degeneration. Application of those cells even increased cell death on day 3 after OGD (p < 0.05) (Fig. 3A).
Influence of OGD-Damaged Hippocampal Tissue on Migration of MNC Fractions
MNCs were labeled with CTO and applied onto mouse slices with transgenic neuronal expression of eYFP (Fig. 4). Fluorescence microscopic evaluation revealed that direct application of MNCs onto damaged tissue caused more cells to adhere to the slice compared to application of the same MNC number on normoxic hippocampal tissue (Fig. 4A vs. B). MNCs remained morphologically unchanged when cocultivated with normoxic control slices, but frequently converted from round to an elongated phenotype when applied onto OGD slices (Fig. 4C). In addition, three-dimensional confocal analyses showed that MNCs preferentially invaded the ischemically damaged nervous tissue (Fig. 4D, E). In contrast, MNCs predominantly remained at the slice surface when applied onto normoxic control slices (Fig. 4D, E vs. F).
Induction of active migration of CD34+ cord blood cells by postischemic hippocampal tissue. (A–F) Only scattered, round MNCs were seen on control slices when 2.5 × 104 Celltracker Orange (CTO)-labeled MNCs (red) were added directly onto hippocampal slices of B6.Cg-TgN, Thy1-eYFP 16 Jrs mice (green fluorescence in pyramidal neurons) after 3 days of cultivation (A). In contrast, numerous MNCs were observed on OGD slices (B). These cells preferentially developed an elongated soma (C, inset in B”). More cells migrated into the tissue (D, E) compared to normoxic control slices (F). CTO signals from MNCs in the slices are indicated by white arrow heads. Scale bars: 100 μm (A, B) and 25 μm in (C, D). (D) Total z view of an OGD slice. Insets in (E) and (F) represent vertical sections at positions indicated by green or red lines. (G–I) Cell-free supernatants were obtained from homogenates of rat hippocampal slices after 3 days of cultivation and added to the lower compartment of a migration chamber. Migration of MNCs towards supernatants was investigated and compared to medium and medium supplemented with stromal cell-derived factor (SDF-1) as negative and positive controls, respectively (G). Hippocampal slices with and without previous OGD induced migration of similar numbers of MNCs (G, p < 0.05 compared to medium control), while strongest migration was observed towards SDF-supplemented medium (G, ***p < 0.001). However, absolute and relative number of migrated CD34+ cells was higher after OGD (H, I) compared to slices maintained under normoxic conditions. Note logarithmic scaling of ordinate in (H).
A migration assay using supernatants of homogenized organotypic cultures (22) was used to obtain information about the cell subsets migrating into lesioned hippocampal tissue. In this assay, normoxic control tissue (undamaged) and tissue that underwent OGD prior to homogenization induced migration of a comparable number of cells (Fig. 4G), which were significantly different from control medium (p < 0.05). In a positive control assay addition of SDF-1 (CXCL-12), a potent chemoattractant for hematopoetic stem and progenitor cells (1), monocytes and lymphocytes (4), markedly induced migration of MNCs towards the lower chamber of the culture system (p < 0.01) (Fig. 4G).
Migrated cells where collected and analyzed by flow cytometry to investigate the chemotactic influence of damaged nervous tissue on different MNC fractions [T cells, B cells, natural killer (NK) cells, CD34+, and CD133+ cells]. Comparison of the total numbers of different migrated cells revealed no obvious differences between OGD slices and normoxic control tissue. In all setups, T cells were the most prominent fraction within migrated MNCs (Fig. 4H, note logarithmic scale). Relative comparison between originally applied cell numbers and migrated cells for all of the investigated cell populations revealed that, in comparison to normoxic tissue (control), OGD-damaged hippocampal tissue preferentially attracted CD34+ stem/progenitor cells (Fig. 4I). However, the difference was not statistically significant. Attraction of other cell fractions did not markedly differ between control and damaged hippocampal tissue extracts.
Effects of Application Mode and Cell Dose on Neurodegeneration in Hippocampal Slice Cultures
MNC application effectively prevented neurodegeneration. We evaluated the role of direct cell contacts for mediation of the protective potential after OGD. For that reason, effectiveness of directly applied MNCs was compared to indirect application. In the latter experiment, direct cell–cell interaction between damaged hippocampal tissue and MNCs was prevented, but exchange of soluble factors between compartments was allowed.
Indirect cocultivation of the same number of MNCs used in the direct cocultivation paradigm (2.5 × 104) had a markedly lower protective effect as revealed by PI staining. In contrast to direct application, this effect could only be discriminated from spontaneous neurodegeneration in untreated OGD slices on day 2 (p < 0.05) (Fig. 5A). To test whether this less pronounced effect could potentially be caused by lowered concentrations of soluble neuroprotective factors after indirect MNC application, rising numbers of cell were applied to OGD-damaged slices (Fig. 5B). Fivefold elevation of MNC number (12.5 × 104) diminished degeneration at all time points investigated following OGD (p < 0.05) (Fig. 5B). On day 2 after OGD, a further increase in the number of MNCs applied (25 × 104) caused an additional significant reduction of PI staining (Fig. 5B), indicating a dose-dependent protection by indirectly applied MNCs.
Dose-dependent reduction of cellular damage after indirect MNC application. For indirect application, MNCs were placed under the slice culture inserts. In contrast to direct application, 2.5 × 104 MNCs were only transiently able to diminish cellular damage in hippocampal slices following OGD (p < 0.05) (A) and failed to provide cellular protection at day 1 (p > 0.4) and day 3 (p = 0.189) (A). Indirect cocultivation using 5- and 10-fold increased cell numbers was able to reduce OGD-induced neural degeneration on all days (p < 0.05). The protective effect of indirect MNC application gradually increased with rising cell concentrations, in particular at days 1 and 2 (B). Error bars show the mean ± SD of three independent experiments each comprising six slices. Values were recorded in relation to OGD at day 1, which was set 100%. *p < 0.05 versus OGD, #p < 0.05 versus 2.5 × 104 MNCs indirect, §p < 0.05 versus 12.5 × 104 MNCs indirect.
Expression of Neurotrophic Factors
Since evaluation of indirect application of MNCs indicated a potential role for soluble protection providing agents, MNC-induced changes in the expression of human and rat derived trophic factors were analyzed. Changes in the expression of human trophic factors secreted by applied MNCs were not detectable because the concentrations of measured human factors (NGF, EGF, IGF-1, BDNF) were below the detection limit of the assay (40 pg/ml).
VEGF secretion from rat hippocampal tissue did not change after direct or indirect application of human MNCs (Fig. 6, left). Evaluation of rat NGF levels revealed a reduction of NGF secretion from OGD slices after application of MNCs in all cocultivation paradigms (Fig. 6, right). Rat CNTF was not secreted in measurable amounts. Because analysis had to be carried out with pooled samples to reach detection limits, descriptive statistics was not applicable.
Expression of neurotrophic factors. Rat neurotrophic factors were measured in supernatants obtained from six rat hippocampal slices that had been cocultivated with human MNCs. Supernatants were pooled for analysis. Bars illustrate relative intensity units. Evaluation of vascular endothelial growth factor (VEGF) secretion from rat hippocampal tissue did not reveal changes, but nerve growth factor (NGF) secretion was reduced in all performed cocultivation setups. Values obtained from normoxic control slices without MNCs were set 100% and are indicated as dashed line.
Discussion
Cell preparations from hUCB have emerged as a highly promising tool for neural tissue protection and promoted regeneration following cerebral ischemia and stroke (24,44). However, the mode of action underlying the therapeutic potential is still a matter of debate. Some investigators showed homing of CD34-enriched hUCB cells to areas bordering the ischemic lesion (8). Stem cells within hUCB MNCs are in principle able to differentiate into neural cells in vitro (12) and in vivo (46), but differentiation from hUCB MNCs into functional neural tissue following stroke remains unproven. Moreover, neuronal differentiation of transplanted cells seems to be a very rare event (8). In contrast, numerous findings indicate that hUCB cells could limit the progress of degeneration by delivering trophic factors (2), by modulation of the neuroinflammatory response (41), and by activation of antiapoptotic pathways (9), thus emphasizing the neuroprotective capabilities of these cells. This study evaluated differences in the protective impact of hUCB cell subpopulations both in vivo and in vitro with emphasis on the role of CD34+ stem cells, direct and indirect cocultivation, and dose-dependent effects to gain more information on relevant cell fractions and their mode of action.
Functional Regeneration and Ischemic Damage Reduction by hUCB Cell Populations
After intravenous transplantation into comorbid rats (38) suffering from focal cerebral ischemia, hUCB MNC, CD34+, and CD34- populations were clearly able to diminish sensorimotor deficits as evaluated in the RotaRod test. Behavioral improvement was detected from day 7 onwards and lasted until the end of the experiment. Findings were confirmed by MRI lesion volumetry and atrophy assessment, with differences between the hUCB cells and controls after 29, but not after 7 days upon stroke onset. To our knowledge, this is the first direct in vivo comparison between hUCB MNCs and subfractions thereof.
Necrotic cell death in the infarct core following cerebral ischemia is an immediate event peaking within 4–12 h following stroke onset (3). Although reduction of necrotic, PI-positive cells in hippocampal slices was shown in this study when hUCB MNCs and CD34+ cells were administered immediately after OGD, cell administration after 24 h in vivo had most likely no effect on necrotic cell death. Any evidence for engraftment of administered cells was not observed in our study 1 month after transplantation. This does not exclude the possibility of cells homing to the brain at early time points following administration. However, reduction of lesion size and brain atrophy most likely cannot be explained by cell replacement or differentiation and might be due to prevention of secondary cell death following MCAO. Inflammatory and apoptotic processes leading to secondary cell loss after stroke have also been observed at later time points following stroke onset in the penumbra and bordering areas (10,43), thus being excellent targets for neuroprotection. Preventing this secondary damage could result in decreased lesion size and brain atrophy even without generation of de novo brain tissue. Neuroprotective abilities of hUCB MNCs were previously shown by our group (15) and recently published studies also reported anti-inflammatory effects of the cells (18,19). Considering further that hUCB MNCs may prevent white matter damage upon ischemia by interactions with the peripheral immune system (14), the mentioned neuroprotective effects could reliably explain both the observed functional improvement and lesion size reduction without detectable cell grafting.
The Role of hUCB CD34+ Cells in Poststroke Neuroprotection
Although to a smaller extent when compared to hUCB MNCs, purified CD34+ cells were able to induce functional recovery and lesion size reduction in MCAO rats but only transiently provided protection in vitro. The data obtained in our study are in partial contrast to the investigation of Taguchi and colleagues (37). This study reported that CD34+ cells from hUCB enhanced neovascularization as well as recruitment and survival of neural progenitor cells following focal cerebral ischemia. This led to cortical regeneration in the model used, a so far unparalleled observation. However, CD34- cells did not have an obvious therapeutic benefit while hUCB MNCs have not been assessed. Albeit the therapeutic impact of hUCB CD34+ was clearly shown in both approaches, there are major differences between the studies. First, the motor defects induced in the Taguchi trial can be considered minor as mice showed rapid recovery from motor deficits, which is in contrast to our approach. This may hinder a direct comparison between both approaches. Next, the Taguchi group used immunodeficient SCID mice. That is an important difference considering the fact that the therapeutic impact of hUCB MNCs and subpopulations may at least partially be related to modulation of inflammatory processes (41) and peripheral immune response following stroke (42). Depletion of CD34+ cells abolished reduction of PI staining after OGD. These results indicate that the diminished neuronal loss after MNC application may be preferentially triggered by CD34+ cells, without CD34+ cells being the most potent neuroprotective cell species itself. The full amount of neuroprotection was only observed after MNC application. A similar phenomenon had been described for purified hUCB CD133+ stem cells, which were able to reduce hypoxia-induced cell death, but not to the extent of hUCB MNCs (27). The relatively small amount of both CD133+ and CD34+ cells among hUCB MNCs also enhances the probability of a rather modulatory role during neuroprotection. It may be assumed that stem and progenitor cells within umbilical cord blood can induce beneficial effects following neuronal damage when acting solely. But for optimal protection maintenance, another, probably mature, cell type might be important. This also points to potential synergistic modes of action between both cell populations, which should be addressed in further studies.
The Therapeutic Role of other hUCB Cell Populations Following Experimental Stroke
Our study revealed that hUCB MNCs are superior to other applied cell populations regarding postischemic neuroprotection in vivo and in vitro. This indicates an important role of other hUCB MNC subpopulations. The main fraction of mature cells within hUCB MNCs are CD4+ T cells (7). Jurkat hT cells used in our study are characterized by expression of CD4 and secretion of interferon-γ (16). hT cells did not induce neuroprotection and even exacerbated tissue damage on day 3 in vitro. In vivo, hT cell-transplanted animals performed inferior to control subjects and tended to develop the largest MCAO-induced lesions. This treatment also came along with largest hemispherical atrophy, although the difference was not statistically significant (p > 0.05). A similar finding was reported for IL-17-producing T-cell subsets (32). These findings clearly argue against hT cells as mediators of neuroprotection. Further studies should address the question whether T-cell depletion can enhance the therapeutic impact of hUCB cell preparations after stroke. Experiments using enriched natural killer or B-cell populations may also be needed for clarification. Interestingly, application of CD34- cells did not provide effective neuroprotection on hippocampal slices after OGD, but resulted in functional improvement and lesion size reduction in vivo. This, at first glance, contradictory result, may point to mechanisms influencing the progress of stroke occurring outside of the brain, which may not require the presence of CD34+ cells. First evidence for such a mechanism was given by Borlongan and coworkers (5). A more recent article reported a significant impact of hUCB MNC transplantation on spleen and splenocyte function that correlated with a reduction in ischemia-induced brain damage (42).
Protective Effects by MNCs After Direct and Indirect Application In Vitro
Two different cell application paradigms were used in the in vitro section of our study: (i) direct application of MNCs and (ii) separate cultivation of hippocampal slices and MNCs to detect neurotrophic effects provided by soluble MNC-derived factors. MNCs indirectly cultivated with neuronal tissue provided less protection to neural tissue when compared with direct application of equal cell numbers. This observation could be explained by concomitant action of soluble and cell contact-based effects. Possibly, diminished beneficial effects were caused by lower concentrations of soluble mediators being present when the cells where placed remote from the postischemic tissue. To test this hypothesis, we applied rising numbers of MNCs in the indirect cultivation setup. Using this approach we were able to achieve protective effects comparable to direct application, but only by using 5- and 10-fold greater cell numbers. This dose dependency further points to the action of soluble factors as one main reason for the reduced neural damage. Moreover, an indirect mode of action strongly suggesting a crucial role of soluble factors has also been shown for bone marrow-derived stem cells (30).
Migration of hUCB Cells After OGD
Migration studies showed that MNCs invaded the hippocampal tissue preferentially after OGD. SDF-1 (CXCL12) as a strong enhancer of cell recruitment was used to demonstrate the migratory potential of MNC fractions. Application of supernatants gained from controls or OGD did not influence the migrating proportion of most cell types with exception of CD34+ cells, which were attracted by ischemic damage of neuronal tissue. This process would contribute to their enrichment at lesioned sites in vivo, a conclusion also drawn by other studies (25,37). Specific accumulation of CD34+ cells would ensure deployment of their observed neuroprotective potential at the right site following systemic application. Posthypoxic and control slice homogenates attracted the same amounts of MNCs in this experiment. This is in contrast to the microscopic investigation, where we detected invading cells in hippocampal slices preferentially after OGD. Most probably, additional release of migratory signals due to tissue damage by homogenization may have accounted for this result. Homing of CD34+ cells was not observed in vivo, as was the case with all other applied cell populations. One explanation could be that the positive sorting of cells influenced their ability to actively migrate over longer distances in vivo, but not over shorter distances as in our in vitro approach. It could also be assumed that a 29-day follow-up period in immunocompetent animals was too long to observe graft survival.
Possible Influence of the Cell Selection Process on Experimental Results
It could be speculated that the positive selection process used in our approach may have limited the neuroprotective potential of CD34+ cells, thus making them unable to act to their full potential. For example, antibodies attached to the cell surface could reduce the migration potential or trigger any kind of intracellular cascades that may have altered the neuroprotective abilities of the cells. However, we did not find signs of cascade activation in a previous study using positively selected CD133+ hUCB cells (27). Positively selected and genetically engineered CD34+ cells were recently shown to have excellent homing abilities and a clear therapeutic impact after transient MCAO (23). Nevertheless, detrimental influences of positive selection cannot be ruled out in general by these observations and need to be addressed in studies especially designed to address this question. For example, positively selected cells could be added back to the MNC fraction. Controlled comparison against another MNC fraction directly obtained from density gradient centrifugation can reveal whether or not migration kinetics and neuroprotective abilities of positively selected hUCB cell populations are altered in the particular disease model. Moreover, as knowledge about potential influences of the selection processes will also be relevant for clinical translation of hUCB-based (or other) cell therapies, there is a clear demand to assess the influence of selection processes in future studies.
Modulation of VEGF and NGF Secretion by hUCB MNCs
Previous results showed a markedly increased VEGF secretion by neuronal cells after hypoxia that was reduced during direct cocultivation of neuronal cells with hUCB MNCs (27). We therefore intended to use VEGF to monitor neuronal response on OGD and MNC application thereafter. Surprisingly, MNC application after OGD did not reduce VEGF secretion by hippocampal cells. Possible explanations of this divergent observation could be glial influences on neuronal VEGF secretion in the organotypic system or a differently regulated release of VEGF from hippocampal glial cells (31).
NGF has been shown to reduce neuronal death in the gerbil hippocampus after ischemia (33). Different from VEGF, secretion of rat NGF was decreased after direct or indirect MNC administration in our experiments. Reduction of cell damage after MNC application could possibly be associated with increased preservation of NGF receptive structures causing the observed decrease of unbound NGF in supernatants. More probably, NGF reduction was due to decreased secretion of NGF by glial cells shown to upregulate NGF expression after ischemia-induced neuronal damage (17). In this case observed NGF reduction would account for reduced neuronal damage and subsequently diminished glial activation after MNC application.
Conclusions
In summary, the experiments described in this study indicate clear differences in neuroprotective capabilities of hUCB cell subpopulations with the whole MNC fraction being the most potent population. CD34+ cells, as a stem cell and progenitor-comprising fraction of hUCB MNCs, preferentially provide neuroprotection at early stages following ischemia. CD34+ cells are able to specifically accumulate at damaged tissue sites, but are not the most potent cell population concerning neuroprotection. MNCs promote neuroprotection at least partly by mechanisms not requiring direct contact to injured neural cells. Moreover, there are indications for a mature cell type within hUCB MNCs influencing stroke progress outside the CNS. This leads to the assumptions of (i) synergistic modes of actions of hUCB MNCs subpopulations and (ii) of a brain-remote component of neuroprotection. Together with the dose-dependent neuroprotective capacity of hUCB MNCs, this concept might also be relevant for therapy translation into the clinic. Application of selected hUCB cell subpopulations does not provide additional therapeutic benefits. Intravenous administration of unselected hUCB MNCs might therefore be a promising therapeutic option for stroke treatment. Time-consuming and difficult to standardize cell selection processes, which potentially also affect cell functionality and viability, and risk-bearing intracerebral delivery might also be unnecessary. In contrast, therapeutic effects most likely depend on sufficient concentrations of protective substances at the site of lesion. Accumulation of systemically applied CD34+ from grafted MNCs at the lesion site could ensure their therapeutic effect while other, possibly mature, cells might more easily reach their sites of action outside the brain.
Footnotes
Acknowledgments
The authors are grateful to Prof. Wilfried Naumann for continuous scientific support during the work and critically reviewing the manuscript. This work was supported by structural founds of the European Union, granted by the Development Bank of Saxony, Dresden, Germany. The authors declare no conflicts of interest. However, please note that during the study, Mrs. Hau was partially employed by VITA34, a private cord blood bank situated in Leipzig.