Primary astrocyte cultures from cerebral cortex were prepared from postnatal 1 day pups as described previously (Wilson and Dixon, 1989). The upper part of the skull was separated and the meningeal tissue removed. The cerebral neocortices were isolated, placed into complete medium, which were containing 10% fetal bovine serum (FΒS), 0.01 M HEPES, 1% penicillin streptomycin in MEM/EΒSS (Minimum Essential Medium with Earle's Balanced Salts, Hyclone). The neocortices were homogenized by pipetting, were plated onto 175 cm2 tissue culture flasks (Corning). The cultures were maintained in a humidified atmosphere of 95% air 5% CO2 at 37℃ for 10~14 days after plating. Flasks were slapped to detach nonadherent cells from the bottom of flasks and then were washed with PΒS. Adherent astrocytes were removed by treatment with trypsin (0.25%, Gibco), were resuspended in complete medium, and suspended cells properly seeded in each condition.
Then microglia and oligodendrocytes were extracted. Once cells reached confluence (5-7 days), they were plated into 100 mm2 (Falcon) at a density of 1.3 X 106 cells/dish. Cells were passaged again and plated. The final percentage of GFAP-expressing cells in these cultures was found to be >95%. When the cells reached confluence, culture medium was changed to serum-free MEM containing 1 % penicillin streptomycin (P/S) and 0.01M HEPES before cytokine treatment. Astrocytes were treated in serum free medium with IL-1β (R&D systems) 10 ng/ml and IFNγ (R&D systems) 10 ng/ml, or TGFβ1 (Peprotech) 10 ng/ml with or
without HGF (Millipore) of various concentrations for 24 hours.
B. Animals and surgical procedures
Adult female Sprague Dawley rats weighing 200-250 g were used in this study. SD rats were housed in the Ajou University Animal Care and Use Committee. For surgery, animals were anaesthetized with chloral hydrate (400mg/kg, i.p.), received a dorsal laminectomy to expose the spinal cord. The dura were opened and iridectomy scissors were used to create a spinal cord right-over hemisected injury at Thoracic 8 level. Vacuum suction was used to aspirate the remaining tissues. Transplants were prepared with the concentration of 4.0 X 104 cells/ul and 5 μl of cell suspensions (total 2.0 X 105 cells) were soaked into gelfoam pledgets.
The gelfoams were immediately implanted into right-hemisection lesions. And then rats by each transplantation groups were injected 4 sites in rostral and caudal of injured spinal cord using a hamilton syringe. Each site was given 1 ul of transplant which is prepared 4 X 104 cells/ul concentrations. So, rats totally transplanted 3.6 X 105 cells / 9ul by concentration.
Experimental groups were divided into control group with phosphate buffered saline (PΒS) soaked gelfoam, MSC only group, and HGF-MSC group with gelfoams soaked with HGF overexpressing MSCs. All animals received daily intraperitoneal cyclosporine (NORVATIS) at a dosage of 10 mg/kg beginning from one day before transplantation to sacrifice.
Prophylactic antibiotics (cefazolin) were intraperitoneally injected on the next day after each surgery, and a bladder care had been provided twice daily until the animals resumed self-voiding.
C. Immunocytochemistry
For immunocytochemistry, cells were plated on 9 mm or 12 mm coverslips coated with poly-D-lysine (Sigma). Attached cells on coverslip were fixed in 4% paraformaldehyde (PFA) for 30min at room temperature (RT) after 3 times PΒS washing each for 10 minutes. After blocking with 10% normal goat serum for 1hr, primary antibodies were applied in the same blocking solution at 4℃ overnight or for 4 hours at RT. After thorough washing with PΒS, appropriate secondary antibodies tagged with Alexa Fluor 488 or Alexa Fluor 594 (Molecular Probes, Eugene, OR) were applied for 1hr at RT. The primary antibodies used in this study were mouse anti-GFAP (Dako; 1:400), rabbit anti-c-met (Santa Cruz; 1:100), rabbit anti-phospho-c-Met (Invitrogen; 1:100), mouse anti-CS-56 (Sigma; 1:400), mouse anti-neurocan (Millipore; 1:100), mouse anti-human mitochondria (Chemicon; 1:400) and rabbit anti-Ki67 (novocasta; 1:400). The coverslips were mounted onto slides with glycerol based mounting medium (Biomeda, Foster City, CA). The images were taken using a FV 300 confocal microscope (Olympus, Tokyo, Japan). These experiments are repeated several trials by experiment number of each group as described the table 1.
D. Tissue processing and immunohistochemistry
Rats were anesthetized with an overdose of chloral hydrate and perfused with heparinized saline (0.9%) followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer. The spinal cord containing lesion site was dissected and post-fixed in 4% PFA for 2 hours, followedby cryoprotection in a graded series of sucrose solutions. Cryosections for
spinal cord (20 μm thickness) were made longitudinally in a 1:10 series, mounted onto Super Frost plus slides (Fisher Scientific, Pittsburgh, PA), and stored at -20℃ until use. For immunohistochemistry, longitudinal tissue sections were treated with 10% normal goat serum (Hyclone, Logan, VT) to prevent nonspecific immunoreactivity. Primary antibodies dissolved in the same blocking solution were applied onto tissue sections at 4℃ overnight followed by appropriate secondary antibodies tagged with Alexa Fluor 488 or Alexa Fluor 594 (Molecular Probes, Eugene, OR) for 1hr at RT. The primary antibodies used for immunohistochemistry were same as those for immunocytochemistry. To quantification of fluorescence intensity, such as GFAP and CS-56, we analyzed confocal images of 3 sites that are rostral, caudal and contra lateral beyond the glial limitance from 3 cryo-section at each group.The images were taken using a FV 300 confocal microscope (Olympus, Tokyo, Japan).
Immunohistochemistry is repeated several trials by experiment number of each group as described the table 4.
E. Enzyme-Linked Immunosorbent Assay (ELISA)
Cultured supernatants or spinal cord lysates were subjected to ELISA analysis to determine the level of TGFβs and HGF. For cultured supernatants, primary astrocyte cultures were treated with cytokines with or without HGF (as described above) for 24 hours and supernatants were collected. Tissue samples were homogenized with a dounce tissue grinder in ice-cold homogenization buffer (50mM Tris-HCl, pH 7.6, containing 150 mM NaCl, 1mM EDTA, 0.32M sucrose) supplemented with 1 homogenized with a douncetor cocktail (Roche,
Mannheim, Germany). The homogenates were then centrifuged at 12,000 rpm for 10 min and the supernatants were used for ELISA analysis. Protein concentration of the cultured supernatants or tissue homogenates were measured using Micro BCA Protein Assay kit (Pierce, Rockford, IL), and equal amounts of samples (typically 10 to 20 µg) were loaded into 96 well plates coated with capture antibodies. Concentration of HGF was assayed using human HGF ELISA kit (immunis, EIA, Japan) and that of TGFβ1 and β2 using kits from R&D systems (MN, USA). Detailed procedures followed an instruction manual provided by corresponding manufacturers. ELISA is tried by experiment number of each group as described the table 2 and table 4.
F. Western Blot analysis
One week after surgery, 17 rats were anesthetized with overdose chloral hydrate and briefly perfused with ice-cold saline to remove blood cells. Five mm-long spinal cord blocks containing lesion site were quickly dissected and homogenized in tissue extraction buffer containing protease inhibitor cocktail (Pierce) and phosphatase inhibitor (Halt Phosphatase Inhibitor Cocktail, Themo), which safeguards against serine, threonine and protein tyrosine phosphatase activities, on ice. Protein concentration was determined using the BCA protein Assay kit (Pierce). Equal amounts (24 ug / 30ul) of proteins were analyzed by SDS-PAGE, together with size marker (Thermo, Rockford, IL). It is performed in 10% polyacrylamide gels with 8% stacking gels. Proteins were separated by constant voltage of 130 mV in RT after separated by 60 mV in stacking gels and transferred to polyvinylidene difluoride
(PVDF) membranes (Immobilon-P; Millipore, Bedford, MA). The membranes were washed in Tris-buffered saline (TΒS-, 0.9% NaCl, 10mM Tris-HCl pH 8.0) containing 0.1% Tween-20 (Sigma) (TΒS-T) and blocked 5% skim milk for 2 hours at RT. The membranes were then incubated with following primary antibodies in 5% ΒSA solution. : anti-GFAP (Dako) and anti-c-met (SantaCruz, 1:1000). Especially anti-phospho-c-met (Invitrogen, 1:1000) were used in 1% ΒSA solution. β-actin was used as a loading control for cell lysates After incubation with horseradish peroxidase conjugated secondary antibodies, immunoreactivity was visualized by chemiluminescence reagents (ECL advance western blot detection kit, GE healthcare).
For neurocan detection, tissue lysates were treated with 0.03 U chondroitinase ABC (protease-free, SEIKAGAKU), which is cut the GAG chain from core protein, for 2 hours at 37℃. SDS-PAGE was performed in 6% polyacrylamide gels with a 4% stacking gel.
Proteins were prepared by 24 ug/30ul concentration and separated by constant voltage of 60 mV in RT after separated by 40 mV in stacking gels. Separated proteins in gel were transferred to PVDF membranes (Millipore) at 4℃ for 15 hours by constant current of 150mA in cold room. The blot membrane was rinsed three times in Tris-buffered saline (TΒS, 0.9% NaCl, 10mM Tris-HCl pH7.5) containing 0.05% Tween 20 (Sigma) (TΒS-T) and blocked 5% skim milk for 2 hours at RT. The blot was incubated with anti-neurocan (Millipore) in 5% skim milk solution. Anti-neurocan antibody can detect the 1D2 (150-163 kDa, C-terminal epitope of neurocan, ), 1F6 (122-130 kDa, N-terminal epitope of neurocan) and intact neurocan (240-270kDa) (Asher et al., 2000). After incubation with horseradish peroxidase conjugated secondary antibodies dissolving in 5% skim milk, immunoreactivity
was visualized by chemiluminescence reagents (ECL advance western blot detection kit, GE healthcare). This analysis is repeated several trials by experiment number of each group as described the table 4
G. Reverse Transcription polymerase chain reaction (RT-PCR)
Total RNA was extracted from cultured cells using Trizol (Gibco) according to the manufacturer’s protocol. The amount of RNA was determined using spectoscopy at 260 nm.
Five ug of RNA was reverse transcribed to cDNA using a standard RT protocol. One ul (0.4 ug) of cDNA was added to PCR-reaction premix (GenDEPOT) with 10 pM corresponding primer pairs. The following primers were used for polymerase chain reaction : GAPDH, 5’-GTG TAG TTC ACG CCC ACG TC-3’ (forward), 5’-5’-GTG ATG GCA TGG ACT 5’-GTG GT-3’
(reverse) ; neurocan, 5’-GCC ACA CTC TAC ACT CGT CCC-3’ (f), 5’TCT CCC CAG CAT AGC CCT GAT-3’(r) ; phosphacan, 5’-GCA AGT CCT GCC GTC CTT GCA 3’-(f), 5’-GGA ATA GGG ATT AGT AAC AGC-3’ (r) ; c-Met, 5’-TGT CTC TGA AAT CCA CCC GA 3’ (f), 5’ GTG TAG TTC ACG CCC ACG TC-3’ (r). Their specificity was verified using the basic local alignment search tool (BLAST) on the GenBank database. PCR amplification was performed with 35 cycles of 95 ℃ for 30 s, 55 ℃ ~ 60 ℃ (property Tm) for 30 s, 72 ℃ for 90 s. The PCR products were separated on a 1% agarose gel and stained by Ethidium Bromide (EtBr). The amount of each product was quantified by a gel document system (Bio-Rad). GAPDH expression was used as an internal reference to verify equal concentrations of cDNA in each sample. We have executed experimental number of each group as described
the table 1. And we quantified mRNA expression level of neurocan and phosphacan by band thickness using the image J program based on band thickness of non-TGFβ1 treatment.
H. Statistical Analysis
Statistical comparison of mean values was performed using one-way ANOVA followed by Tukey’s post hoc tests. All values are expressed as mean ± SD. Quantitation graphs were generated by GraphPad Prism version 4.00 (GraphPad Software, San Diego, CA, USA)
Ⅲ. RESULTS
A. HGF prevents cytokine-induced astrocytic activation in vitro.
To examine whether HGF affects formation of glial scars, we used primary cultured astrocytes isolated from 1 day pub rat’s neocortex to mimic glial scars (Fig. 1A). When more than 95% cultured astrocytes express glial fibrillary acidic protein (GFAP), we induced astrocyte activation by treatment of TGFβ1 at a concentration of 10 ng/ml (Baghdassarian et al., 1993; Reilly et al., 1998) (Fig. 1B, C). Immunocytochemistry revealed that TGFβ1-treated astrocytes upregulated GFAP positive intermediate filaments and showed hypertrophic morphology. HGF was added to the medium at three different concentrations: 5 ng/ml, 50 ng/ml, and 250 ng/ml. Fifty ng/ml and 250 ng/ml treatment of HGF obviously prevented morphological change, although 5 ng/ml was not effective (Fig. 1E-G). To quantify changes in astrocytic morphologies, we measured mean GFAP positive areas (Fig.
1G). We found that treatment of TGFβ1 clearly increased the mean size of astrocytes and co-treatment of HGF at 50 ng/ml and 250 ng/ml significantly reduced the extent of hypertrophic changes.
It has been previously reported that HGF stimulated neuronal cell survival and oligodendrocyte precursor cell proliferation (Yan and Rivkees, 2002; Ohya et al., 2007). To examine whether HGF affects proliferation of astrocytes, cells were stained with anti Ki-67, which is a cellular marker of proliferation and detects nuclei-during interphase. Ki-67 immunoreactivity was not different between different treatment groups (Fig. 2A-C).
Quantification of the number of Ki-67 positive cells showed that proliferation between TGFβ1 treatment and HGF treatment with TGFβ1 was not significantly different (Fig. 2E).
As shown in Fig 2D, DAPI counts in regular area made no difference between TGFβ1 and HGF treatment, including control.
We next examined expression of HGF receptor c-Met in primary astrocyte. RT-PCR showed that c-Met mRNA is present in primary astrocytes and its expression is greatly increased after TGFβ1 treatment (Fig. 3A). HGF treatment did not change expression of c-Met. By immunostaining with antibodies that detect the beta-chain of c-Met (Fig. 3B-D), we confirmed the expression of c-Met in HGF treated astrocytes.
Fig. 1. Astrocytic activation in vitro glial scars model A: Illustration of primary astrocyte cultures and experimental scheme. B: GFAP stained astrocytes in serum free medium. B: 10 ng/ml treatment of TGFβ1 induced astrocytic hypertrophy. C: 5ng/ml treatment of HGF. D-F:5 ng/ml, 50 ng/ml, and 250 ng/ml HGF was added to culture medium with TGFβ1 respectively. Cells were treated with the above conditions for 24 hours and were fixed for immunocytochemistry with GFAP. G: Quantification of GFAP positive areas. *p<0.05,
**p<0.01 and ***p<0.001 by one-way ANOVA followed by Tukey’s post hoc test for the comparison of mean GFAP areas between different groups.
Fig. 2. Absence of HGF effect on astrocyte proliferation A-C: Ki-67 immunocytochemistry and DAPI staining of primary astrocytes without TGFβ1 (A), with TGFβ1 (B), and TGFβ1 with HGF 50 ng/ml (C). D: DAPI counting results by groups. E:
Quantification of Ki-67 positive cells counts among the all strocytes in the one taken picture.
Fig. 3. Expression of HGF receptor, c-Met, in primary astrocytes A: RT-PCR analysis showed mRNA expression of c-Met in primary astrocytes with or without TGFβ1 and HGF contreatment condition. B-D: Images of immunocytochemistry with c-Met and DAPI in primary astrocytes treated with TGFβ1 and HGF 50 ng/ml. B: Immunoreactivity of DAPI. C. Immunoreactivity of c-Met. D: merge image of DAPI and c-Met immunoreactivities.
B. HGF decreases expression of chondroitin sulfate proteoglycans (CSPGs)
To examine whether HGF affects production of CSPGs that is a major component of glial scars, we measured mRNA expression of neurocan and phosphacan, which are two species of CSPGs upregulated following cytokine stimulation (McKeon et al., 1999; Asher et al., 2000) and spinal cord injury (Jones and Tuszynski, 2002). TGFβ1 induced astrocytic activation resulted in dramatic increases in neurocan and phosphacan mRNAs.
Quantification data showed that neurocan mRNA expression was elevated 8 times higher than control level. HGF treatment at a concentration of 50 ng/ml almost completely blocked the increase of neurocan mRNA expression (Fig. 4A, B). Phosphacan mRNA level was also increased 2 times with TGFβ1, and HGF co-treatment evidently reduce the expression level (Fig. 4C, D). These experiments confirmed that HGF obviously suppresses mRNA production of CSPGs.
Fig. 4. Modulation of chondroitin sulfate proteoglycans (CSPGs) mRNA expression by HGF A: Quantification of mRNA expression of neurocan. B: neurocan RNA analysis by RT-PCR. C: Quantification of mRNA expression of phosphacan D: phosphacan RNA analysis by RT-PCR. *p<0.05, **p<0.01 , and ***p<0.001 by one-way ANOVA followed by Tukey’s post hoc test for the comparison with hypertrophic effects in primary astrocytes.
Table 1. Experimental groups and treatment number in vitro
C. Effects of HGF on TGFβ secretion
Following spinal cord injury, TGFβ1 and TGFβ2 were secreted from reactive astrocytes around the lesion site. TGFβ1 induces inflammatory responses and is involved in initial formation of the glial scars, while TGFβ2 maintains glial scars at later time points (Buss et al., 2008). Inhibition of TGFβ1 and TGFβ2 functions by neutralizing antibodies reduced the extent of scar formation (Logan et al., 1999; Moon and Fawcett, 2001). Therefore, TGFβs are regarded as strong inducers for glial scars formation.
For these reasons, we determined whether reduction of astrocytic activation is associated with changes in secretion of TGFβs. To this end, we adopted a different activation scheme. Treatment of interleukin-1 beta 1 (IL-1β) and interferon-gamma (IFNγ) has been shown to strongly induce astrocytic activation (Hewett et al., 1993). Applying IL-1β and IFNγ together is known to increase GFAP expression by NO induction (Brahmachari et al., 2006).
Purified astrocytes were treated with 10 ng/ml of IL-1β and 10 ng/ml of IFNγ for 48 hours after starvation during 24 hours (Fig. 5A). Treatment of IL-1β and IFNγ together elicited dramatic changes in astrocytic morphology (Fig 5B-D). Astrocytes that were exposed to IL-1β and IFNγ showed highly elongated morphology and tended to aggregate together compared to untreated astrocytes (Fig 5C). HGF treatment partially restored astrocytic morphology (Fig 5E). To compare secretion level of TGFβ1 and TGFβ2 from astrocytes cultured supernatants were harvested and analyzed for TGFβ1 and TGFβ2 ELISA after 48 hours of cytokines addition (Fig 5A, F-G). TGFβ1 secretion was significantly increased by IL-1β and IFNγ, which was completely blocked by HGF co treatment (Fig 5D).
TGFβ2 secretion is not significantly increased by addition of IL-1β and IFNγ, but clearly decreased by HGFs (Fig 5E). Thus, HGF inhibited secretion of TGFβ1 and TGFβ2 in IL-1β / IFNγ induced activated astrocyte culture system.
Fig. 5. HGF inhibited secretion of TGFβ1 and TGFβ2 from activated astrocytes. A:
Schematic illustration of the experimental procedure. B-D: Representative images of GFAP Immunocytochemistry of primary astrocyte without cytokine treatment (B), with IL-1β and IFNγ at 10 ng/ml each for 48 hours (C), and addition of 50 ng/ml of HGF with IL-1β and IFNγ treatment (D). E-F: ELISA analysis of TGFβ1 (E) and TGFβ2 (F) in supernatants collected from cultures for 48 hours under designated experimental conditions. *p<0.05, and ***p<0.001 by one-way ANOVA followed by Tukey’s post hoc test for the comparison of TGFβ1 and TGFβ2 concentration between different groups.
Table 2. Experimental groups and treatment number of IL-1β and IFNγ-induced in vitro scar model
ELISA (TGFβ1 and TGFβ2)
groups number
Non IL-1β and IFNγ (control) 4
IL-1β 10 ng/ml + IFNγ 10 ng/ml 4
IL-1β 10 ng/ml + IFNγ 10 ng/ml + HGF 5 ng/ml 4
D. Ex vivo delivery of HGF using mesenchymal stem cells in hemisection spinal cord injury
In vitro data demonstrated that HGF prevents cytokine-induced astrocytic hypertrophy
and reduced CSPGs expression. We attempted to test whether HGF could regulate glial scars formation in vivo. To achieve stable and robust delivery of HGF, we transplanted HGF overexpressing-mesenchymal stem cells (MSCs) into the hemisection spinal cord injury site at T8 level (Fig. 6A). After hemisection spinal injury, we immediately implanted gelfoams soaked with PΒS (control group), human bone marrow derived MSCs (MSC alone), and MSCs transduced with HGF gene (HGF-MSC).
Immunohistochemistry was performed to detect surviving grafted cell at 2 weeks after surgery. GFP immunoreactivity was used to detect MSCs because enhanced GFPs were incorporated into MSCs. MSCs were also detected by human specific mitochondria antibodies. Spinal cord tissue of control group (PBS) did not show any immunoreactivity for eGFP or human mitochondria (Fig. 6B). In animals with MSCs alone or HGF-MSCs, eGFP and human mitochondria positive grafted cells were detected inside gelfoam as well as spinal cord parenchyme surrounding the lesion (Fig 6C-F). Thus, grafted MSCs were found to survive for at least 2 weeks post transplantation in both gelfoam around circumference of lesion.
We next examined if HGF overexpressing MSCs were capable of secreting HGF outside of cells in vitro. ELISA analysis of supernatants obtained from culture for different duration showed that the level of HGF secretion increased as the duration of culture was prolonged (Table 3). In contrast, the amount of HGF in supernatants from MSC alone cultures was
almost negligible. To verify adequate delivery of HGF using the ex vivo approach, animals were sacrificed 1 week after injury, and the level of HGF was measured in the spinal cord tissue around the lesion site. The level of HGF in the spinal cord with HGF-MSCs was 14 times higher than that of control or MSC group (Fig. 7A). This result indicated that the ex vivo delivery of HGF using MSCs successfully delivered HGF to the injured spinal cord
tissue. Immunohistochemistry using anti-HGF antibodies was also performed to detect presence of HGF in the injured spinal cord tissue (Fig 7C-E). Confocal image showed diffuse HGF immunoreactivity surrounding hemisected injury site in HGF-MSCs group (Fig 7E). Partial expression of PΒS and MSCs groups might be due to endogenous expression of HGFs (Fig. 7C, D).
Binding of HGF to the c-Met receptor tyrosine kinase (RTK) triggers receptor dimerization, which is originally disulfide linked α-β heterodimeric RTK, and phosphorylation on multiple residues (Ma et al., 2003). To determine whether c-Met signaling was effectively induced by the ex vivo approach, phosphorylation of c-Met was measured using phospho c-Met specific antibodies can detect phosphorylation of Tyr 1230, 1234, and 1235 (Fig. 7B). Expression of c-Met receptor was detected to a similar extent in all three groups. Apparent phosphorylation of c-Met was observed in HGF-MSC group, and c-Met phosphorylation was also detected in MSC alone group to a much smaller extent (Fig.
7B).
Fig. 6. Ex vivo delivery of HGF gene using human mesenchymal stem cells as carriers A:
Fig. 6. Ex vivo delivery of HGF gene using human mesenchymal stem cells as carriers A: