Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Stem Cells. Author manuscript; available in PMC 2012 September 18.
Published in final edited form as:
PMCID: PMC3445430

Human Embryonic Stem Cell-Derived Oligodendrocyte Progenitor Cell Transplants Improve Recovery after Cervical Spinal Cord Injury


Evidence that cell transplants can improve recovery outcomes in spinal cord injury (SCI) models substantiates treatment strategies involving cell replacement for humans with SCI. Most pre-clinical studies of cell replacement in SCI examine thoracic injury models. However, as most human injuries occur at the cervical level, it is critical to assess potential treatments in cervical injury models and examine their effectiveness using at-level histological and functional measures. To directly address cervical SCI, we used a C5 midline contusion injury model and assessed the efficacy of a candidate therapeutic for thoracic SCI in this cervical model. The contusion generates reproducible, bilateral movement and histological deficits, although a number of injury parameters such as acute severity of injury, affected gray to white matter ratio, extent of endogenous remyelination, and at-level locomotion deficits do not correspond with these parameters in thoracic SCI. Based on reported benefits in thoracic SCI, we transplanted human embryonic stem cell (hESC)-derived oligodendrocyte progenitor cells (OPCs) into this cervical model. hESC-derived OPC transplants attenuated lesion pathogenesis and improved recovery of forelimb function. Histological effects of transplantation included robust white and gray matter sparing at the injury epicenter, and in particular, preservation of motor neurons that correlated with movement recovery. These findings further our understanding of the histopathology and functional outcomes of cervical SCI, define potential therapeutic targets, and support the use of these cells as a treatment for cervical SCI.

Keywords: spinal cord injury, human embryonic stem cells, forelimb, pathogenesis, neuroprotection


The pathogenesis of contusive SCI involves a complex process that begins with cord compression, immediate axon and cell damage, hemorrhage, and hypoperfusion 1. This trauma initiates a secondary degenerative cascade that exacerbates the loss of neurons, oligodendrocytes and myelin, as well as axons 2, 3. Concomitant to these secondary processes and cell loss are inflammation and immune responses 4, the proliferation of progenitor cells 57, migration of astrocytes to the injury 8, and gliosis and cyst formation 9, 10. Thus, treatment of this multifactorial injury will likely require a combination therapy capable of addressing disparate injury components.

More than half of all human spinal cord injuries occur at the cervical level, with C4–C6 injuries accounting for nearly 40% of all cases 11. As in thoracic injuries, cervical injuries interrupt axonal tract conduction between the cranial and caudal central nervous system. Notable differences in cervical relative to thoracic cord include enlarged neuronal pools, proximity of descending axons to their cell bodies, and at-level neuronal mediation of limb movement. In persons with SCI, injury level changes the priorities of recovery targets between the tetraplegic population and the paraplegic population; most notably, that recovery of hand and arm function is ranked as the priority of the tetraplegic population 12. These differences highlight an increased clinical relevance for studies of cervical SCI models that examine the direct relation of at-level histological effects and motor function. The C5-level is the most common level at which to induce bilateral 1315 and unilateral 16 cervical injuries. Anatomical evidence of cervical innervations indicates that, in general, C2–C5 level motor neurons project to muscles of the shoulder and proximal forelimb and C6–C8 level motor neurons project to distal forelimb muscles 17. Accordingly, injuries to C5 are demonstrated to affect both at-level and below-level outcomes such as locomotion and digit function, respectively. A broader understanding of the histopathology and functional outcomes of cervical SCI could hasten the identification of appropriate therapeutic targets for this injury and support the translation of potential therapeutics to the sizable cervical SCI population.

Cell replacement strategies are inherently combination therapies, as they confer phenotype-specific benefits as well as neurotrophic benefits to surrounding tissue. We have previously shown that transplantation of hESC-derived OPCs 18 to thoracic SCI resulted in pathotropism, cell survival and differentiation, enhanced remyelination, and improved locomotor outcomes 19 without harmful effects 20. Other studies identified a number of neurotrophic factors and chemokines expressed by hESC-derived OPCs, and demonstrated that the secreted factors enhanced neuronal survival and neurite outgrowth 21, 22. Thus, hESC-derived OPCs may confer benefit by pleiotropic effects such as myelinating demyelinated axons or providing neurotrophic support to surrounding tissue. In the present experiment, we explored the ability of hESC-derived OPCs to provide anatomical and functional benefit following transplantation into acute cervical SCIs.


Spinal cord injury

Cervical spinal cord contusion injuries were performed on female Sprague Dawley adult rats (200–220 g) as previously described 14. A dorsal laminectomy was performed on the fifth cervical vertebra (C5) to expose the spinal cord, and the rat was suspended by clamps on the vertebrae cranial and caudal to the laminectomy. Contusion injury was induced using the Infinite Horizon Impactor (Precision Systems, Kentucky, IL) with a force of 200 kdyn (n = 46). After contusion, the deep and superficial muscle layers were sutured, and the skin was closed with stainless-steel wound clips. Immediately after surgery, animals were given subcutaneous saline and prophylactic Baytril (2.5 mg/kg/d, s.c.; Bayer, Shawnee Mission, KS) and maintained on an isothermic pad until alert and mobile. Animals received manual bladder expression twice daily and were inspected for weight loss, dehydration, and distress, with appropriate veterinary care as needed.

Derivation of OPCs from hESCs

The WA07 (H7) hESC line at passage 32 was obtained from Geron (Menlo Park, CA). Cells were expanded in hESC growth media 23 and differentiated according to published protocols 18, 24, 25. Briefly, dissociated colonies were placed into low attachment flasks in 50% hESC growth media and 50% glial restriction media (GRM) for 2 days. On day 1 the media was 4 ng/ml basic FGF, 20 ng/ml EGF, and on day 2 was supplemented with EGF and 10 μM all-trans-retinoic acid (RA) (Sigma-Aldrich). This media was then replaced with 100% GRM supplemented with EGF and RA for an additional 7 days. Cells were then exposed for 25 days to GRM/EGF without RA. At day 28, yellow spheres were plated in flasks coated with 1:30 Matrigel for 7 days. Cultures were then trypsinized, remnant spheres excluded, and cells replated on Matrigel and cultured for 7 days in GRM/EGF. The total time for the differentiation protocol was 42 days.

For immunocytochemistry, cells were plated on slides coated with poly-L-lysine and human laminin (Sigma-Aldrich). For transplantation, cells were concentrated to 200,000 cells/μl. Trypan blue exclusion testing indicated that this preparation consisted of 87–98% viable cells at time of transplant.

Behavioral testing

Forelimb movement scores were determined from videotape of animals crossing a clear Plexiglas walkway marked with 1 cm grid lines on the floor. Prior to testing, each animal was acclimated to the apparatus and trained to cross for intermittent food reward. During test trials, animals were videotaped from underneath (used for forelimb stride length) and from the side (used for proximal forelimb step range and passed-perpendicular step frequency) of the walkway using a Canon NTSC digital video camcorder (ZR10; Canon, Tokyo, Japan). Videos were analyzed at the middle of a pass, frame by frame, using media player software. All behavioral tests and assessments were conducted in a blinded manner (n=13/group).

For forelimb stride length, animals were videotaped from below the walkway each week for weeks 1 through 3, and then every other week for weeks 5 through 9. The videos were scored for forelimb stride length as outlined for hindlimb analysis by Gonzalez et alias 26. Briefly, forelimb stride length was defined as distance from the start of a step with the forepaw through to the end of that step with the same paw. Measurements were taken on each side for five steps and averaged for each animal.

Measures of proximal forelimb included proximal forelimb step range and passed-perpendicular step frequency. These measures were selected to correspond with reported C5 motor neuron innervation of proximal forelimb by McKenna et alias 17. These measures were recorded from videotape taken from the side view during week 9. For the step range parameter, the range of proximal forelimb movement, expressed as an angle, was determined. The angle of proximal forelimb movement was defined as the inside angle of the proximal forelimb at the placement of a step to the inside angle at the lift off of that same step. Measurements were taken on the left and right sides for three steps and averaged for each animal. For the passed-perpendicular step frequency, the proximal forelimb positions at the lift-off of each of five steps were determined for the left and right sides. The position of the forelimb at this point was compared to a line perpendicular (90°) to the floor and aligned at the shoulder of the animal. The number of steps that the proximal forelimb crossed the line perpendicular to the floor were counted as passed-perpendicular. The occurrence of steps passed-perpendicular relative to the total number of steps was averaged for each animal.


Both groups received cyclosporine A (20 mg/kg/d, s.c.; Bedford Laboratories, Bedford, OH) 1 day prior to transplantation/vehicle administration and then everyday for the duration of the study. Cell transplantation or vehicle (control) administration occurred 7 days after contusion injury. Animals were anesthetized as above, and the laminectomy site was re-exposed. After immobilization of the spinal process cranial to the contusion site, a 10 μl Hamilton syringe (Hamilton, Reno, NV) was lowered into the spinal cord using a stereotactic manipulator arm. Cell suspensions were injected along the midline of the spinal cord at a depth of 1.2 mm into one site cranial and one site caudal to the lesion epicenter, in a total volume of 7.5 μl (1,500,000 cells) at a rate of 1 μl/min. Control animals received an equal volume of vehicle only at the same injection rate. The needle was removed after 5 min.

hESC-derived OPC gene expression profile

Total RNA was isolated from cultured OPCs using Trizol reagent (Invitrogen) followed by DNAse I digestion (Ambion, Austin, TX) and clean up with RNeasy (Qiagen, Valencia, CA) according to manufacturers’ instructions. Reverse transcription (RT) of mRNAs was performed using M-MLV Reverse Transcriptase (Ambion) with random hexamers and poly-dT as primers. cDNAs generated by RT were used for subsequent 30-cycle PCRs using a Mastercycler thermal cycler (Eppendorf), Platinum Taq DNA polymerase (Invitrogen) and specific primers for: TGFβ1 (forward: 5′-agagatacgcaggtgcaggt-3′, reverse: 5′-tcagagttgcactccgaaga-3′), PDGFRα: (forward: 5′-tgtgtgggacattcattgct-3′, reverse: 5′-gggtactgccagctcacttc-3′), Neuregulin1 (forward: 5′-caaagaaggcagaggcaaag-3′, reverse: 5′-aactggtttcacaccgaagg-3′), Neuregulin2 (forward: 5′-ggagaccagagaccgcctac-3′, reverse: 5′-taaaaacgcctttgccgtta-3′), and Nkx2.2 (forward: 5′-agcctacatttctgcgtgct-3′, reverse: 5′-gcctcacttggtcaattcgt-3′) (Invitrogen). PCR products were analyzed by gel electrophoresis.


Animals were killed 8 weeks after cell transplantation under pentobarbitone anesthesia by aortic perfusion with isotonic, heparinized saline followed with 4% paraformaldehyde (Fisher Scientific, Pittsburgh, PA) in 0.1 M phosphate buffer, pH 7.4. The spinal cord was divided into eight 1 mm blocks that extended 4 mm cranial to and 4 mm caudal to injury epicenter. Alternate blocks were processed to produce resin or cryostat sections. Resin sections were used to determine the number of remyelinated axons, the gross pathology of the transplant environment, and morphometric measurements. Cryostat sections were used for immunodetection and hematoxylin and eosin (H&E) stain.

For resin processing, blocks were postfixed for 24 h in 4% glutaraldehyde (Fisher Scientific) and embedded in resin (Electron Microscopy Sciences) according to standard protocols. Transverse semithin (1 μm) sections were cut from the cranial face, stained with alkaline toluidine blue, coverslipped, and examined by light microscopy on an Olympus AX-80 microscope and a 2 megapixel MagnaFire digital camera using Olympus MicroSuite B3SV software (Olympus America, Melville, NY). For electron microscopy, blocks were trimmed and sections were cut at 100 nm, mounted on copper grids, uranyl acetate and lead citrate stained, and viewed under a Hitachi EM 600 electron microscope at 75 kV.


Antibody detection was performed on fixed cultured cells or cryosectioned spinal cord using standard protocols, as previously described 19. Antibodies used included Oct-4 (rabbit, 1:500), O4 (mouse, 1:50), NG2 (rabbit, 1:100), Olig1 (rabbit, 1:200), A2B5 (mouse, 1:100), PDGFRα (rabbit, 1:200), RIP (mouse, 1:200) mouse anti-human nuclei (1:30) (all from Millipore, Temecula, CA), class III β-tubulin (mouse, Tuj1; 1:200), Pax6 (rabbit, 1:100) (all from Covance Research Products, Denver, PA), cow GFAP (rabbit, 1:500; DakoCytomation, Glostrup, Denmark), mouse anti-APC/CC1 (1:200; Calbiochem, San Diego, CA), mouse anti-human CXCR4 (1:200 Abcam, Cambridge, UK), rabbit anti-rat/mouse CXCL12 (eBioscience, San Diego, CA) and stage-specific embryonic antigen 4 (SSEA4) supernatant (mouse, 1:5; gift from Geron Corporation, Menlo Park, CA).

Cell and sections were imaged using an Olympus AX-80 microscope (Olympus America, Melville, NY) and MagnaFire digital camera with the Olympus MicroSuite B3SV software. For transplant cultures, the percentage of immunopositive cells was determined by dividing the total number of immunopositive cells by the total number of Hoechst-positive cells in each imaging chamber and then averaging the results from three chambers per marker. To determine distribution of transplanted cells, the number of human nuclei immunopositive cells were counted on three sections 100 μm apart from each tissue block for each animal and averaged; the counts within corresponding blocks from animals within a group were then used for statistical comparison. Double labeling of cells with anti-human nuclei plus cellular differentiation markers was confirmed using confocal microscopy (MRC 1000; Bio-Rad, Hercules, CA; Zeiss, Thornwood, NY). Single confocal plane images of human nuclei and phenotype markers were collected and combined to produce three-dimensional reconstruction and determine label colocalization. Cells immunopositive for Olig1 or APC/CC1 were quantified from three fields within the region(s) of interest per section. The Olig1 or APC/CC1 counts were averaged from three sections for each animal and the averages used for statistics. Motor neuron survival was determined from the average of motor neuron counts taken from five transverse sections 100 μm apart and centered on the injury epicenter. The motor neuron counts for each section were averaged for each animal. The averaged counts were used for statistical comparison of motor neuron preservation across treatment groups.

Myelin pathology counts

To quantify normally myelinated, demyelinated, and oligodendrocyte or Schwann cell (SC)-remyelinated axons, regions of pathology were located on 1 μm resin sections at 40x magnification and traced using the Olympus MicroSuite B3SV software to calculate perimeter and area as previously reported 14. Normally myelinated, demyelinated, and oligodendrocyte or SC-remyelinated axons were counted using the line-sampling technique detailed by Blight 27. The average number of normally myelinated, demyelinated, and oligodendrocyte or SC-remyelinated axons within 5 25 μm2 × 25 μm2 areas along the radial line yielded an estimate of the total number of axons within a region of pathology, and is calculated as the number of axons per square millimeter. At least 10% of the area of pathology is used determine total axons for each area of pathology. The number of oligodendrocyte- remyelinated, SC-remyelinated, and demyelinated axons was used to determine the oligodendrocyte remyelination efficiency, calculated as the ratio of oligodendrocyte remyelinated axons to SC-remyelinated and demyelinated axons.


Morphometric analysis was performed as previously described 14. Breifly, 1 μm transverse semithin spinal cord sections were imaged at 40x using an Olympus AX-80 microscope and MagnaFire digital camera and measurements were traced on the image using the Olympus MicroSuite B3SV software to determine area and perimeter values. The accuracy of delineation was checked by viewing delineations at 200x and 1000x magnification. Spared white matter area was defined as white matter area(s) marked by less than 50% pathology that included aberrant hallmarks such as swelling, damaged axons, hypercellularity, and gross demyelination. Multiple areas of pathology within a section were individually traced and then summed. Spared gray matter area was defined as gray matter area(s) that were clearly identifiable from white matter and cavitation of transplant. The measured perimeters per section were used to derive the value of maximal area for each parameter as detailed by Schrimsher and Reier 15. Maximal area was used for comparisons between animals to account for potential tissue processing artifacts. Maximal area is defined as the area of circle calculated from the perimeter of the area measured. The maximal areas were averaged from animals within each group.

Quantitative, differential gene expression profile in spinal cord

Differential expression was assessed on epicenter segments of C5-injured, adult female rats at days 14 and 21 post-injury and compared to uninjured rats (n=4/group). These time points reflect the gene expression profile of untreated spinal cord at days 7 and 14 post-transplantation, respectively (n=4/group).

Total RNA was isolated from homogenized rat spinal cord using Trizol reagent (Invitrogen) and a Sonic Dismembrator (Fisher Scientific). Total RNA was DNAse treated with Turbo DNAfree reagent (Ambion) and cleaned with an RNeasy kit (Qiagen) according to manufacturers’ instructions. Reverse transcription (RT) of mRNAs was performed using M-MLV Reverse Transcriptase (Ambion) with random hexamers and poly-dT as primers. cDNAs generated by RT were used for subsequent 40-cycle Real-Time PCRs using a Mastercycler Realplex thermal cycler (Eppendorf) and a SensiMix NoRef kit (Quantance/Bioline). Mastermix included 200nM concentration of specific primers for GAPDH (forward: 5′-atgactctacccacggcaag -3′, reverse: 5′-acgccagtagactccacgac -3′), HGF (forward: 5′-tggctgtacaatccctgaaa -3′, reverse: 5′-gagctactcgtaataaaccatctgc -3′), TGF2β (forward: 5′-atgaacctttcattgcccttg -3′, reverse: 5′-gctcagttctataacggctcaca -3′), Caspase 4 (forward: 5′-tctccaaactcatttcctgctt -3′, reverse: 5′-gccttttcaaatgattgttgc -3′), GADD45 (forward: 5′-gcttcctccttcagtctcacc -3′, reverse: 5′-acgccagtagactccacgac -3′), Pycard (forward: 5′-caacacaggcaagcactcat -3′, reverse: 5′-caggctggagcaaagctaaa -3′), Fas (forward: 5′-tgattgcatctcgtttgtgg -3′, reverse: 5′-tgcagcctgtaagtgatatttga -3′), TNRFSF1A (forward: 5′-accaagtgccacaaaggaac -3′, reverse: 5′-ctggaaatgcgtctcactca -3′), TNFRSF1B (forward: 5′-catgtcaacgtcacctgcat -3′, reverse: 5′-ctgggactgagagggacact -3′), CRP (forward: 5′-ttcgtatttcccggagtgtc -3′, reverse: 5′-tctcgttaaagctcgtcttgg -3′), CD40 (forward: 5′-tctgagccctggaactgttt -3′, reverse: 5′-tattactgcggacccctgac -3′), IL10 (forward: 5′-cctgctcttactggctggag -3′, reverse: 5′-tgtccagctggtccttcttt -3′), NeuN (forward: 5′-tggagcaaatctcccagttc -3′, reverse: 5′-atagcagccatcatccttgg -3′), and ChAT (forward: 5′-gaggagcagttcaggaagagcc -3′, reverse: 5′-agatgaggctggctgcaaacc -3′) and 100ng/μl concentration of cDNA. Primers were designed using NCBI or Primer3 websites. Quantitative PCR (qPCR) data was confirmed by melt curve plot and analyzed by the comparative CT method 28.

Statistical methods

Forelimb stride length scores were analyzed using two-way, repeated measures ANOVA with Tukey’s multiple comparison test at each time point. The treatment group (transplant vs. control) was set as the between-groups factor and the week interval was set as the within-group, repeated-measures factor. The SPSS 11.5 t test was used to determine differences between proximal forelimb range scores for transplanted and control groups. The SPSS 11.5 t test was used to determine differences between quantitative histological and qPCR values.


Directed Differentiation of OPCs from hESCs

Undifferentiated hESCs expressed SSEA4- and Oct-4 (data not shown). At the end of the 42-day differentiation protocol, cells had a bipolar morphology characteristic of immature OPCs and a typical antigenic profile; they were immunopositive for the OPC markers Olig1 (>80%), NG2 (>90%) (Fig. 1a) and PDGFRα (>70%) (Fig. 1b). A small number of A2B5-positive cells (<5%), GFAP-positive astrocytes (<1%) and Tuj1-positive neuronal cells (<5%) were also identified. No cells were detected within the transplant population that were labeled with the more mature oligodendrocyte markers O4 and RIP (Fig. 1c). Also absent were cells that labeled with the hESC markers Oct-4 and SSEA4 or the neural progenitor marker Pax6 (Fig. 1c).

Figure 1
hESC-derived OPC expression profile. (a): More than 90% of cells labeled with the chondroitin sulfate proteoglycan NG2 (red), associated with oligodendrocyte precursors. (b): A total of 72 ± 2% of cells expressed the cell surface tyrosine kinase ...

hESC-derived OPCs were additionally profiled by RT-PCR (Fig. 1d). Confirmation of PDGFRα mRNA expression supports the antibody detection of this receptor on OPCs (Fig. 1b). Selected cDNAs for the secreted proteins TGFβ1, Neuregulin 1, and Neuregulin 2 and for the transcription factor Nkx2.2 were present in detectible quantities.

Bilateral Cervical Contusion Pathology

Contusion produced bilateral forelimb impairment, and did not result in respiratory distress or failure in any animals, indicative of an overall functional preservation of more cranial phrenic motor neuron pools. All animals regained bladder function and were capable of hindlimb locomotion and self-access to food and water within 7-days post-injury.

Examination of the non-transplanted spinal cords 9 weeks after injury revealed bilateral cavitation centered at the impact site that extended approximately 2 mm along the cranial-caudal neuraxis. Examination of transverse planes of the cord revealed a maximal area of cavitation at the injury epicenter (Fig. 2a). Morphometric analysis at this site revealed that the maximal area of the cavity comprised approximately 18% of the total area of the spinal cord. The injured spinal cord showed marked contusion pathology that included gray matter loss, axonopathy, scattered demyelination, oligodendrocyte remyelination, partitioned Schwann cell (SC)-remyelination (Fig. 2b), inflammatory infiltrates, and perivascular cuffing (Fig. 2c).

Figure 2
Anatomical features of midline, bilateral cervical contusion. (a): Transverse, toluidine blue-stained section at the injury epicenter depicts the loss of gray matter and throughout the white matter, with concentrations of pathology in the medial ventral ...

Transplanted OPCs Survive, Localize to Injury, and Differentiate

Transplanted hESC-derived OPCs survived and localized to the injury site during the nine week study period (Fig. 3). OPCs were detected by anti-human nuclear staining and were present in all transplanted animals (Fig. 3a), confirming xenograft survival. Animals without transplants did not exhibit such labeling. At the injury epicenter, anti-human positive cells were identified throughout the transverse plane, but were concentrated around the area of former cavitation (Fig. 3a). Rodent cells were also present within this area, as identified by methyl green-only staining (Fig. 3b). Quantitation of human nuclei revealed the highest number of cells at the injury epicenter and very few cells cranial and caudal to the epicenter (Fig. 3c).

Figure 3
Transplantation of hESC-derived OPCs into acute SCI resulted in cell survival, limited migration from the site of implantation, and differentiation to mature oligodendrocytes. (a): The distribution of hESC-derived OPCs was detected using anti-human-positive ...

The ability of transplanted OPCs to retain their phenotype and mature post-transplantation was examined by double immunostaining for human-nuclei and the oligodendrocyte-specific markers Olig1 or adenomatous polyposis coli tumor suppressor protein (APC/CC1). Cells that colabeled with human nuclei and Olig1 antibodies comprised 94%±5% of the total number of human nuclei positive cells (Fig. 3d). Cells that colabeled with human nuclei and APC/CC1 comprised 9%±4% of the total number of human nuclei positive cells. In sections 2–3 mm away from the injection site and cranial to the injury epicenter, hESC-derived OPCs localized to the dorsal column. Human specific anti-CXCR4 labeled transplanted cells were found juxtaposed to mouse/rat specific CXCL12-positive loci in these same areas (Fig. 3e). In the white matter, APC/CC1 cells constituted a much larger percentage of the human-nuclei positive cells, comprising 66%±10% of the total human nuclei positive cell population (Fig. 3f), indicating that the transplanted OPCs became oligodendrocytes in regions appropriate for myelinogenic differentiation. Cells double-labeled with human nuclei and APC/CC1 processes were confirmed with reconstruction of confocal thin-plane scans (Fig. 3g).

OPC Transplantation Improves Forelimb Motor Function After Injury

To assess the effects of transplanted OPCs on recovery of forelimb function after injury, we examined forelimb stride length as a general gait parameter. Because C5 innervation impacts proximal forelimb function, we also analyzed proximal forelimb step range and passed-perpendicular step frequency.

Assessment of forelimb locomotion on the forelimb stride length task was significant (p < 0.05) on two-way, repeated measures analysis. The difference in the outcome measure between transplanted and non-transplanted groups became significant (p < 0.05) by 4 weeks post-transplant, and the transplanted group showed significantly (p < 0.001) longer stride length than controls for the duration of the study (Fig. 4a). In addition, stride length within the control group peaked at 1 week post-transplant whereas stride length within the transplant group peaked at 6 weeks post-transplant. Thus, the transplanted group demonstrated more extensive recovery before reaching a functional plateau. As a reference, the mean stride length of uninjured animals that were trained to the same apparatus for 9 weeks was 17.5 cm (dashed line).

Figure 4
hESC-derived OPC-transplants to cervical SCI significantly improved forelimb motor function. (a): Recovery of forelimb mean stride length significantly improved in the transplanted group relative to the non-transplanted group. The improvement was significant ...

Examination of the proximal forelimb range of motion exhibited during stepping for each group showed a significant (p < 0.01) increase in step range of the transplanted group as compared to the non-transplanted group (Fig. 4b). The increased step range observed in the transplanted group occurred at the lift-off of the step and not at the placement, as determined by a significant (p < 0.01) difference of the inside-angle between the forelimb and the floor at lift-off and no significant (p > 0.05) difference in the inside-angle at placement (Fig. 4c). Notably, the mean lift-off angle of the control group was about 90°, or perpendicular to the floor. Stepping function was also analyzed by determining the frequency of steps that passed the perpendicular plane. The occurrence of passed-perpendicular steps was significantly (p < 0.001) more frequent in the transplanted group as compared to the non-transplanted group (Fig. 4d). This measure is consistent with a classification of frequent (>50%) in transplants versus occasional (≤50%) in controls. Together these findings suggest that forelimb locomotion and range of motion are improved by OPC transplants.

OPC Transplantation Alters Lesion Pathogenesis

The injury epicenter within non-transplanted spinal cords exhibited widespread loss of white matter and gray matter (Fig. 5a). Cavity borders lacked cells and axons, consistent with glial scar formation (Fig. 5c). White matter pathology was most evident in the dorsal column and medial ventral white matter (Fig. 5a). Non-transplanted spinal cords contained extensive oligodendrocyte remyelination and partitioned SC-remyelination with scattered demyelination (Fig. 5e). In addition, perivascular cuffing and inflammatory infiltrates were present in the control animals, suggestive of dynamic, ongoing pathology within these focal lesions.

Figure 5
Histopathogenesis in transplanted and non-transplanted animals. (a): Microthin sections from non-transplanted animals demonstrate myelin pathology concentrated around the dorsal and medial-ventral white matter and cavitation that eliminates a large portion ...

The injury epicenter within transplanted spinal cords was characterized by broad white matter and gray matter sparing (Fig. 5b). Most notably, the transplanted spinal cords lacked cavitation (Fig. 5b) and a distinct lesion border (Fig. 5d). The transplant area contained a high concentration of OPCs with a homogeneous distribution that suggests a lack of astrocyte partitioning. As in control spinal cords, the white matter of transplanted cords contained oligodendrocyte-remyelinated axons and scattered demyelinated axons. In contrast to the non-transplanted spinal cords, transplanted spinal cords contained more normally-myelinated axons and fewer SC-remyelinated and demyelinated axons (Fig. 5f).

Quantification of normally myelinated, oligodendrocyte-remyelinated, SC-remyelinated, and demyelinated axons within the dorsal column exhibited significantly (p < 0.001) more normally myelinated axons and significantly fewer SC-remyelinated (p < 0.001) and demyelinated (p < 0.05) axons in the transplanted group compared to the non-transplanted group (Fig. 5g). Because the number of oligodendrocyte-remyelinated axons is limited by the number of SC-remyelinated and demyelinated axons, the ratio of oligodendrocyte-remyelinated to SC-remyelinated and demyelinated axons was determined in order to assess the efficiency of oligodendrocyte-remyelination. The non-transplanted animals had an average oligodendrocyte-remyelination efficiency of 1.13±0.31. The transplant group had a significantly (p < 0.001) higher oligodendrocyte-remyelination efficiency of 7.64±0.75 (Fig. 5h). The transplant group, therefore, had an oligodendrocyte-remyelination efficiency that was approximately 680% that of the control group.

The characteristic injury-induced cavitation within non-transplanted spinal cords was absent within transplanted spinal cords. Morphometric analyses at the injury epicenter showed no significant differences (p > 0.05) between the transplanted and non-transplanted groups in the spinal cord maximal area (Fig. 6a), but a significant (p < 0.05) reduction of cavitation in the transplant group (Fig. 6b). This reduction in cavity was evident throughout the 2 mm extent of the lesion (data not shown). In addition, quantification of spared cord at the injury epicenter revealed significant increases in the total spared white matter area (p < 0.01) (Fig. 6c) and total spared gray matter area (p < 0.001) (Fig. 6d) in the transplants compared to controls. Further analysis of gray matter sparing revealed a significant (p < 0.05) (Fig. 6e) increase in ventral gray matter (VGM) sparing and a significant (p < 0.05) (Fig. 6f) increase in average spared motor neurons within the spared VGM.

Figure 6
Injury epicenter morphometry and motor neuron counts in transplanted and non-transplanted animals. (a): Morphometric analysis of the injury epicenter revealed no significant difference in maximal area. (b): Cavitation was significantly decreased in the ...

To test for correlation between functional and histological outcomes, the test of proximal forelimb range of motion was compared to several histological outcome measures. A significant correlation was found for proximal forelimb range of motion and mean motor neuron sparing (r = .7153, p < 0.01), and ventral gray matter sparing (r = .6612, p < 0.05), suggesting that the reduction in motor neuron loss and gray matter loss may account for the improved locomotion in the transplant group. Together, these data suggest that hESC-derived OPC transplantation can attenuate tissue loss and thereby preserve spinal cord mediated function.

Transplantation of OPCs Affects Gene Expression in Acute Phase of Spinal Cord Injury

To elucidate the acute events that precede hESC-derived OPC transplantation-induced sparing, we examined the differential expression of select genes during the acute phase after SCI. The select genes of interest (GOIs), listed in the Supplemental Table, were detectible in injured spinal cord samples. Fold changes in expression at day 21 were detected as significantly (* p < 0.05, ** p < 0.01, *** p < 0.001) different in select genes for neurotrophic factors, apoptosis, inflammation, and a neuronal cell marker (Fig. 7a).

Figure 7
Differential expression of select genes of interest (GOIs) at injury epicenter for neurotrophic factors, apoptosis, inflammation, and neurons in transplanted compared to non-transplanted animals. (a) Time course of differential expression of select GOIs ...

To assess the ability of hESC-derived OPC transplants to affect SCI-induced gene expression patterns, differential gene expression was examined in transplanted and non-transplanted animals at 21 days post-injury (14 days post-transplant). At this timepoint, expression of relevant, specific genes deviated from the injured pattern and tended toward an uninjured, or spared, pattern. Specifically, fold difference expression of genes for HGF (p < 0.05), IL10 (p < 0.001), and ChAT (p < 0.05) were significantly increased, whereas genes for Casp4 (p < 0.001), GADD45 (p < 0.01), Pycard (p < 0.05), Fas (p < 0.01), TNFR1a (p < 0.01), TNFR1b (p < 0.01), CRP (p < 0.01), and CD40 (p < 0.05) were significantly decreased (Fig. 7b). These results indicate that hESC-derived OPC transplantation can affect acute SCI-induced gene expression changes consistent with sparing and support the histological and functional findings at 8-weeks post-transplantation.


This study directly assessed the potential for hESC-derived OPCs to provide at-level benefit to cervical SCI. Our results complement other reports of cervical transplantation research 29 and suggest that outcome from treatment can differ from that in the thoracic spinal cord. These findings further our understanding of the pathology and functional outcomes of cervical SCI, and are the first to demonstrate anatomical and functional benefit following transplantation of a hESC-derivate to cervical SCI.

Cell-based therapeutics have proven successful in pre-clinical SCI models 30 at least in part due to their ability to address multiple features of SCI such as cell loss, demyelination, or homeostatic loss. Several cell replacement strategies have emerged to treat SCI, including O-2A progenitors 31, 32, Schwann cells 29, 33, 34, or neural stem cells 35, 36. Several recent studies indicate that myelinogenic transplants elicit histologic repair and functional recovery following SCI 36, validating demyelination as a therapeutic target for SCI 14, 3739. Nistor et alias 18 and subsequently Izreal et alias 40 described protocols to direct the differentiation of hESCs into high purity OPC populations, and demonstrated their myelination potential. The use of hESCs as a source for human transplant populations offers advantages over other cell types, including the inherently broad capacity for expansion and differentiation 41.

The ability of non-myelinating cells such as bone marrow stromal cells to improve the outcome of SCI exemplifies therapeutic benefits of a non-myelinogenic transplant 42, 43. In an analogous manner, hESC-derived OPCs might contribute to at-level histological and functional outcomes via a non-myelin mechanism. A recent report identified 49 neurotrophic factors expressed by hESC-derived OPCs 22, including the neurotrophic factors insulin-like growth factor 1, brain-derived neurotrophic factor, NT-3, nerve growth factor, and transforming growth factor-β1. Co-cultures of hESC-derived OPCs with cortical neurons enhanced neurite outgrowth from the cortical neurons, indicating that the secreted factors can affect surrounding cells 21, 22. These results suggest that hESC-derived OPCs might produce beneficial effects in SCI aside from remyelination, such as neuroprotection, suppression of inflammation, promotion of axonal regeneration, and/or homeostatic maintenance.

Our methods for cervical contusion produced cavitation and pathological features consistent with moderate to severe bilateral injuries reported by Schrimsher and Reier 15 and Pearse et al. 13. Here, the lesion emanated a distance of less than 2mm from the epicenter and quantitative analysis revealed extensive remyelination. This is in contrast to thoracic injuries of the same force, which resulted in lesions extending 6–12mm either side of the lesion epicenter 14 with substantial demyelination and remyelination. This difference in remyelination is consistent with results of Franklin et alias 44 that found the migratory potential of endogenous myelinogenic cells is restricted to approximately 2 mm either side of a region of demyelination. The robust endogenous remyelination in this cervical model might also reflect enhanced axonal survival as a result of the axotomies being relatively closer to the cell bodies of origin 45. Such potential asymmetries between cervical and thoracic injuries underscore the need to test the effectiveness of potential treatments for SCI in both injury models.

Most of the transplanted hESC-derived OPCs localized to the lesion epicenter. However, some transplanted cells were found up to 2 mm away from the site of injection. The expression of CXCR4 by hESC-derived OPCs and the detection of human-CXCR4 positive cells within areas of rat-CXCL12 reactivity outside of the injury epicenter suggest that hESC-derived OPCs can follow migratory cues. These findings cannot rule out the possibility that dispersion during the transplant injection can also contribute to the cell distribution. Quantification of Olig1 and APC/CC1 positive cells in regions of high concentrations of human nuclei-positive cells within injury epicenter suggests that a lack of niche prevents maturation of a high percentage of the transplanted cells. This feature is supported by evidence that within the injured white matter niche, the percentage of APC/CC1 and human nuclei co-labeled cells is increased. The persistent detection of Olig1 or APC/CC1 on human cells localized to the injury epicenter suggests that transplanted hESC-derived OPCs retained an oligodendroglial lineage despite the derivation method and diverse molecular milieu. These data, together with the absence of human cells expressing astrocyte, neuronal, or embryonic markers, suggests that the transplant population did not trans-differentiation or become pluripotent.

Although transplanted hESC-derived OPCs differentiated into mature oligodendrocytes within white matter tracts, no significant difference in remyelination was detected between the transplanted and non-transplanted groups. This result is not unexpected given the extensive endogenous remyelination evidenced in the control group. Importantly, the treatment group had significantly more normally myelinated axons and fewer Schwann cell and demyelinated axons relative to control, suggesting that fewer axons were demyelinated in the treatment group. This would potentially present less substrate for remyelination and thereby reduce the number of remyelinated axons in the treatment group. Comparison of oligodendrocyte remyelination efficiency supports this interpretation, as the treatment group showed an increased efficiency relative to control.

The forelimb outcome measures used in this study to determine locomotor recovery were based on the results of anatomical tracing of C5 afferents to forelimb musculature by McKenna et alias 17. In this cervical contusion model, animals achieved a high level of recovery of the affected limbs at an earlier timepoint, as compared to that reported for thoracic contusion animals 19. The transplant group had an increase of stride length, proximal forelimb range of motion, specifically in the step lift-off, and a relative increase in frequency of passed-perpendicular steps. Thus, the improvement in forelimb function in the transplant group was observed across multiple forelimb gait parameters relevant to the level of injury.

Morphometry and cell quantification analyses revealed a number of significantly improved histological outcomes in the transplants compared to controls. Correlation of the different histological outcomes with a measure of proximal forelimb function in both transplants and non-transplants indicated a statistically significant correlation of forelimb function with spared motor neurons and gray matter sparing, underscoring the importance of preserving at-level motor neuron function in cervical SCI. Interestingly, no significant correlations were present when forelimb functional outcomes were compared to maximal cavitation or maximal cord areas, indicating that the reduction in cavitation alone was insufficient to account for the improved locomotion in the transplant group. Although this is consistent with hESC-derived OPC-mediated neuroprotection of cultured neurons21, 22, the differential gene expression results demonstrate that transplantation of OPCs reduce markers of SCI-induced apoptosis and inflammation, and support neuron survival in vivo during the acute phase after injury. Having identified significant gene expression changes at 21 days post-injury, we expect it will be possible to further investigate these pathways to form a better understanding how hESC-derived OPCs interact or interfere with inflammation and cell death mechanism, such as TNFα–induced excitotoxicity 46 or microglial activation 47, to promote tissue sparing.

These findings demonstrate that transplantation of human OPCs into acute cervical SCI improves histological outcomes that correlate with improved recovery. Importantly, our data indicate that cervical SCI presents a distinct lesion pathogenesis and that the mechanism and outcome of treatment can differ from that in the thoracic spinal cord. These findings underscore the importance of using cervical injury models in addition to thoracic models for the preclinical development of therapeutics for SCI, in order to better address the human SCI population.

Supplementary Material

Supp Table S1


This work was supported by the Geron Corporation, the University of California Discovery Grant, the Roman Reed Spinal Cord Injury Research Fund of California, Research for Cure, and individual donations to the Reeve-Irvine Research Center. J.S. and M.S. were supported by the Bill and Joan Jackson Scholarship. We thank Jane Lebkowski, Catherine Priest, Scott Thies, and Edward Wirth for discussion and advice. We thank Sharyn Rossi, Saba Motakef, Adrian Tripp, and Audrey Keebaugh for assistance with animal care, Sarah Park and Stephen Marley for data collection, and Adriana Gutierrez and David Ferguson for assistance with cell culture. HSK is chairman of the scientific advisory board of California Stem Cells, Inc. GN is a member of the scientific advisory board of California Stem Cells, Inc.


Author Contributions:

Jason Sharp: Conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing

Jennifer Frame: Collection and/or assembly of data, manuscript writing

Monica Siegenthaler: Collection and/or assembly of data, data analysis and interpretation, manuscript writing

Gabriel Nistor: Provision of study material or patients, collection and/or assembly of data

Hans S. Keirstead: Conception and design, financial support, administrative support, data analysis and interpretation, manuscript writing, final approval of manuscript


1. Hagg T, Oudega M. Degenerative and spontaneous regenerative processes after spinal cord injury. Journal of neurotrauma. 2006;23:264–280. [PubMed]
2. Crowe MJ, Bresnahan JC, Shuman SL, et al. Apoptosis and delayed degeneration after spinal cord injury in rats and monkeys. Nature medicine. 1997;3:73–76. [PubMed]
3. Kakulas BA. A review of the neuropathology of human spinal cord injury with emphasis on special features. The journal of spinal cord medicine. 1999;22:119–124. [PubMed]
4. Ankeny DP, Popovich PG. Mechanisms and implications of adaptive immune responses after traumatic spinal cord injury. Neuroscience. 2008 [PMC free article] [PubMed]
5. Horky LL, Galimi F, Gage FH, et al. Fate of endogenous stem/progenitor cells following spinal cord injury. The Journal of comparative neurology. 2006;498:525–538. [PMC free article] [PubMed]
6. Mothe AJ, Tator CH. Proliferation, migration, and differentiation of endogenous ependymal region stem/progenitor cells following minimal spinal cord injury in the adult rat. Neuroscience. 2005;131:177–187. [PubMed]
7. Yamamoto S, Yamamoto N, Kitamura T, et al. Proliferation of parenchymal neural progenitors in response to injury in the adult rat spinal cord. Experimental neurology. 2001;172:115–127. [PubMed]
8. Frisen J, Johansson CB, Torok C, et al. Rapid, widespread, and longlasting induction of nestin contributes to the generation of glial scar tissue after CNS injury. The Journal of cell biology. 1995;131:453–464. [PMC free article] [PubMed]
9. Faulkner JR, Herrmann JE, Woo MJ, et al. Reactive astrocytes protect tissue and preserve function after spinal cord injury. J Neurosci. 2004;24:2143–2155. [PubMed]
10. Silver J, Miller JH. Regeneration beyond the glial scar. Nature reviews. 2004;5:146–156. [PubMed]
11. NSCISC. Facts and Figures at a Glance. Available at:
12. Anderson KD. Targeting recovery: priorities of the spinal cord-injured population. Journal of neurotrauma. 2004;21:1371–1383. [PubMed]
13. Pearse DD, Lo TP, Jr, Cho KS, et al. Histopathological and behavioral characterization of a novel cervical spinal cord displacement contusion injury in the rat. Journal of neurotrauma. 2005;22:680–702. [PubMed]
14. Siegenthaler MM, Tu MK, Keirstead HS. The extent of myelin pathology differs following contusion and transection spinal cord injury. Journal of neurotrauma. 2007;24:1631–1646. [PubMed]
15. Schrimsher GW, Reier PJ. Forelimb motor performance following cervical spinal cord contusion injury in the rat. Experimental neurology. 1992;117:287–298. [PubMed]
16. Gensel JC, Tovar CA, Hamers FP, et al. Behavioral and histological characterization of unilateral cervical spinal cord contusion injury in rats. Journal of neurotrauma. 2006;23:36–54. [PubMed]
17. McKenna JE, Prusky GT, Whishaw IQ. Cervical motoneuron topography reflects the proximodistal organization of muscles and movements of the rat forelimb: a retrograde carbocyanine dye analysis. The Journal of comparative neurology. 2000;419:286–296. [PubMed]
18. Nistor GI, Totoiu MO, Haque N, et al. Human embryonic stem cells differentiate into oligodendrocytes in high purity and myelinate after spinal cord transplantation. Glia. 2005;49:385–396. [PubMed]
19. Keirstead HS, Nistor G, Bernal G, et al. Human embryonic stem cell-derived oligodendrocyte progenitor cell transplants remyelinate and restore locomotion after spinal cord injury. J Neurosci. 2005;25:4694–4705. [PubMed]
20. Cloutier F, Siegenthaler MM, Nistor G, et al. Transplantation of human embryonic stem cell-derived oligodendrocyte progenitors into rat spinal cord injuries does not cause harm. Regenerative medicine. 2006;1:469–479. [PubMed]
21. Faulkner J, Keirstead HS. Human embryonic stem cell-derived oligodendrocyte progenitors for the treatment of spinal cord injury. Transplant immunology. 2005;15:131–142. [PubMed]
22. Zhang YW, Denham J, Thies RS. Oligodendrocyte progenitor cells derived from human embryonic stem cells express neurotrophic factors. Stem cells and development. 2006;15:943–952. [PubMed]
23. Carpenter MK, Inokuma MS, Denham J, et al. Enrichment of neurons and neural precursors from human embryonic stem cells. Experimental neurology. 2001;172:383–397. [PubMed]
24. Hatch MN, Nistor GI, Keirstead HS. Oligodendrocyte Differentiation from Human Embryonic Stem Cells. In: Loring J, Wesselschmidt RL, Schwartz PH, editors. Human Stem Cell Manual: A Laboratory Guide. 1. New York: Elsevier; 2007. pp. 210–226.
25. Hatch MN, Nistor G, Keirstead HS. Derivation of high-purity oligodendroglial progenitors. Methods Mol Biol. 2009;549:59–75. [PubMed]
26. Gonzalez R, Glaser J, Liu MT, et al. Reducing inflammation decreases secondary degeneration and functional deficit after spinal cord injury. Experimental neurology. 2003;184:456–463. [PubMed]
27. Blight AR. Cellular morphology of chronic spinal cord injury in the cat: analysis of myelinated axons by line-sampling. Neuroscience. 1983;10:521–543. [PubMed]
28. Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparative C(T) method. Nat Protoc. 2008;3:1101–1108. [PubMed]
29. Schaal SM, Kitay BM, Cho KS, et al. Schwann cell transplantation improves reticulospinal axon growth and forelimb strength after severe cervical spinal cord contusion. Cell transplantation. 2007;16:207–228. [PubMed]
30. Eftekharpour E, Karimi-Abdolrezaee S, Fehlings MG. Current status of experimental cell replacement approaches to spinal cord injury. Neurosurgical focus. 2008;24:E19. [PubMed]
31. Groves AK, Barnett SC, Franklin RJ, et al. Repair of demyelinated lesions by transplantation of purified O-2A progenitor cells. Nature. 1993;362:453–455. [PubMed]
32. Rosenbluth J, Schiff R, Liang WL, et al. Xenotransplantation of transgenic oligodendrocyte-lineage cells into spinal cord-injured adult rats. Experimental neurology. 1997;147:172–182. [PubMed]
33. Takami T, Oudega M, Bates ML, et al. Schwann cell but not olfactory ensheathing glia transplants improve hindlimb locomotor performance in the moderately contused adult rat thoracic spinal cord. J Neurosci. 2002;22:6670–6681. [PubMed]
34. Martin D, Robe P, Franzen R, et al. Effects of Schwann cell transplantation in a contusion model of rat spinal cord injury. Journal of neuroscience research. 1996;45:588–597. [PubMed]
35. Cummings BJ, Uchida N, Tamaki SJ, et al. Human neural stem cells differentiate and promote locomotor recovery in spinal cord-injured mice. Proceedings of the National Academy of Sciences of the United States of America. 2005;102:14069–14074. [PubMed]
36. Karimi-Abdolrezaee S, Eftekharpour E, Wang J, et al. Delayed transplantation of adult neural precursor cells promotes remyelination and functional neurological recovery after spinal cord injury. J Neurosci. 2006;26:3377–3389. [PubMed]
37. Guest J, Herrera LP, Qian T. Rapid recovery of segmental neurological function in a tetraplegic patient following transplantation of fetal olfactory bulb-derived cells. Spinal Cord. 2006;44:135–142. [PubMed]
38. Klussmann S, Martin-Villalba A. Molecular targets in spinal cord injury. Journal of molecular medicine (Berlin, Germany) 2005;83:657–671. [PubMed]
39. Totoiu MO, Keirstead HS. Spinal cord injury is accompanied by chronic progressive demyelination. The Journal of comparative neurology. 2005;486:373–383. [PubMed]
40. Izrael M, Zhang P, Kaufman R, et al. Human oligodendrocytes derived from embryonic stem cells: Effect of noggin on phenotypic differentiation in vitro and on myelination in vivo. Molecular and cellular neurosciences. 2007;34:310–323. [PubMed]
41. Sharp J, Keirstead HS. Therapeutic applications of oligodendrocyte precursors derived from human embryonic stem cells. Current opinion in biotechnology. 2007;18:434–440. [PubMed]
42. Sheth RN, Manzano G, Li X, et al. Transplantation of human bone marrow-derived stromal cells into the contused spinal cord of nude rats. J Neurosurg Spine. 2008;8:153–162. [PubMed]
43. Hofstetter CP, Schwarz EJ, Hess D, et al. Marrow stromal cells form guiding strands in the injured spinal cord and promote recovery. Proceedings of the National Academy of Sciences of the United States of America. 2002;99:2199–2204. [PubMed]
44. Franklin RJ, Gilson JM, Blakemore WF. Local recruitment of remyelinating cells in the repair of demyelination in the central nervous system. Journal of neuroscience research. 1997;50:337–344. [PubMed]
45. Fernandes KJ, Fan DP, Tsui BJ, et al. Influence of the axotomy to cell body distance in rat rubrospinal and spinal motoneurons: differential regulation of GAP-43, tubulins, and neurofilament-M. The Journal of comparative neurology. 1999;414:495–510. [PubMed]
46. Hermann GE, Rogers RC, Bresnahan JC, et al. Tumor necrosis factor-alpha induces cFOS and strongly potentiates glutamate-mediated cell death in the rat spinal cord. Neurobiology of disease. 2001;8:590–599. [PubMed]
47. Nicholas R, Stevens S, Wing M, et al. Oligodendroglial-derived stress signals recruit microglia in vitro. Neuroreport. 2003;14:1001–1005. [PubMed]