Conceived and designed the experiments: SJAD MN MM-P JED CP. Performed the experiments: SJAD C-HS JED CP. Analyzed the data: SJAD JED CP. Contributed reagents/materials/analysis tools: SJAD MM-P JED CP. Wrote the paper: SJAD MN JED CP.
The authors have declared that no competing interests exist.
Repairing trauma to the central nervous system by replacement of glial support cells is an increasingly attractive therapeutic strategy. We have focused on the less-studied replacement of astrocytes, the major support cell in the central nervous system, by generating astrocytes from embryonic human glial precursor cells using two different astrocyte differentiation inducing factors. The resulting astrocytes differed in expression of multiple proteins thought to either promote or inhibit central nervous system homeostasis and regeneration. When transplanted into acute transection injuries of the adult rat spinal cord, astrocytes generated by exposing human glial precursor cells to bone morphogenetic protein promoted significant recovery of volitional foot placement, axonal growth and notably robust increases in neuronal survival in multiple spinal cord laminae. In marked contrast, human glial precursor cells and astrocytes generated from these cells by exposure to ciliary neurotrophic factor both failed to promote significant behavioral recovery or similarly robust neuronal survival and support of axon growth at sites of injury. Our studies thus demonstrate functional differences between human astrocyte populations and suggest that pre-differentiation of precursor cells into a specific astrocyte subtype is required to optimize astrocyte replacement therapies. To our knowledge, this study is the first to show functional differences in ability to promote repair of the injured adult central nervous system between two distinct subtypes of human astrocytes derived from a common fetal glial precursor population. These findings are consistent with our previous studies of transplanting specific subtypes of rodent glial precursor derived astrocytes into sites of spinal cord injury, and indicate a remarkable conservation from rat to human of functional differences between astrocyte subtypes. In addition, our studies provide a specific population of human astrocytes that appears to be particularly suitable for further development towards clinical application in treating the traumatically injured or diseased human central nervous system.
The recognition that astrocyte dysfunction may play an important role in a wide range
of neurological disorders raises the question of whether astrocyte transplantation
could be of therapeutic value in treating the injured or diseased human central
nervous system (CNS). For example, it has long been known that astrocytes within
glial scar tissue contribute to the failure of axon regeneration across sites of
traumatic brain or spinal cord injury
There are a number of challenges inherent in the development of astrocyte-based
treatments for human disease. One of the most important of these is the question of
whether all astrocytes are equivalent in their ability to promote repair, or whether
specific populations of astrocytes are more useful than others. While previous
studies had demonstrated a synergistic effect of BMP and LIF on the astrocytic
differentiation of human neural stem cells
We now show that astrocytes generated from the same population of human fetal glial precursor cells, by exposure to either bone morphogenetic protein (BMP) or ciliary neurotrophic factor (CNTF), promote widely divergent outcomes with respect to repairing the injured adult spinal cord. Transplantation of astrocytes generated by exposure of human glial progenitor cells (hGPCs) to BMP (hGDAsBMP) promoted robust behavioral recovery and multi-laminae protection of spinal cord neurons following spinal cord injury (SCI), while transplantation of undifferentiated hGPCs or astrocytes generated by hGPC exposure to CNTF (hGDAsCNTF) failed to provide such benefits. These results provide a defined population of human astrocytes suitable for further pre-clinical development for treatment of SCI, and demonstrate that pre-differentiation into astrocytes prior to transplantation provides a much greater functional recovery than transplantation of precursor cells themselves. Our results also underscore the importance of function-based analysis of astrocyte diversity as a foundation for the development of astrocyte transplantation-based therapies.
As a first step towards determining whether human glial progenitor cells (hGPCs)
can generate functionally distinct astrocyte populations, we exposed embryonic
hGPCs isolated from spinal cords of 9.5 week old abortuses to BMP or CNTF. Both
BMP and CNTF-induced astrocyte populations express GFAP (
Both BMP and CNTF promote the differentiation of hGPCs into GFAP-positive astrocytes expressing S100b but with otherwise distinct morphological and antigenic phenotypes. hGDAsBMP express lower levels of GFAP and exhibit a more compact morphology. hGDAsCNTF have a more elongated morphology and expressed high levels of GFAP. hGDAsCNTF also expressed high levels of neurite-outgrowth inhibitory chondroitin sulfate proteoglycans, phosphacan and CSPG4, as well as the transcription factor Olig2 - all of which have been found to be upregulated in glial scar associated astrocytes. (A–C) Human GPCs grown in bFGF (A) were induced to differentiate into astrocytes using BMP-4 (B) or CNTF (C). Labeling with anti-GFAP (Alexa-488) demonstrates that both BMP4 and CNTF induce differentiation of human glial precursors into GFAP-expressing astrocytes, while Olig2 expression (Alexa-568) is repressed in hGDAsBMP. Scale bar = 50 µm. (D) RT-QPCR analysis of hGPC, hGDABMP and hGDACNTF populations reveals induction of AQP4 and S100β in both hGDAsBMP and hGDAsCNTF. Induction of CX43, GLT1, AKAP12 and GDNF however are restricted to hGDAsBMP. Average fold change and SD of expression levels is shown for three independent experiments using 9W-1 hGPCs. (E and F) Phosphacan and CSPG4 remain elevated in hGDAsCNTF and are reduced in hGDAsBMP derived from both 9W-1 (E) and 9W-2 (F) glial precursors. Mean relative protein expression and SD from three independent experiments are shown. Values were normalized to β-actin and expression in hGPCs.
To test the functional properties of these distinct astrocyte populations in vivo, hGDAsBMP, hGDAsCNTF or undifferentiated hGPCs were transplanted into the injury site of adult Sprague-Dawley rats that had received unilateral transections of the right-side dorso-lateral funiculus (DLF), including the rubrospinal pathway, at the C3/C4 intervertebral spinal cord level.
Serial section analysis of transplants using antibodies to human mitochondria
(hMito) showed that the majority of hGDABMP transplants (4 out of 6)
and half of the hGDACNTF transplants (3 out of 6) that under went
histological analysis had robust survival of hMito+ cells at 5 weeks post
transplantation within dorsolateral funiculus (DLF) injury sites. Surviving
transplants spanned the rostral to caudal extent of injury sites to effectively
provide continuous substrates for potential growth of host axons across sites of
injury. Qualitative assessment of transplant size showed that all surviving
hGDACNTF transplants were larger than hGDABMP
transplants in terms of both their rostral to caudal and lateral to medial
extents (
Immuno-staining for human mitochondrial marker (red channel) of
histological cross sections at sites of injury revealed
hGDAsBMP (A, C, E) and hGDAsCNTF (B, D, F)
transplant masses spanning the dorsal-ventral and lateral-medial margins
of injury sites. Arrowheads in C and D indicate accumulations of
hGDAsBMP and hGDAsCNTF respectively at the
pial surface of lateral funiculus white matter (see also
We next found that hGDABMP grafted injury sites exhibited higher
densities of axons than hGDACNTF grafted injury sites. As shown in
Analysis of the migration of transplanted hMito+ hGDAs revealed similar
patterns of distribution within spinal cord gray matter. The highest densities
of both types of hGDAs were observed within laminae 4, 5 and 6 of gray matter
directly adjacent to sites of injury (
High power imaging of the centers of both types of hGDA grafts showed comparable
densities of hMito positive cell bodies and processes that contained GFAP+
intermediate filaments (
Images of hMito+ (red) and GFAP (green) immuno-reactivity within hGDAsBMP (A, C, E, F) and hGDAsCNTF (B, D, G, H) transplants at the center of injury sites showing comparable densities of GFAP+ intermediate filaments (green) within hMito immuno-positive cell bodies and process of both types of hGDAs. Arrowheads indicate some examples of GFAP+/hMito+ hGDA cell bodies at low and high magnification. Images are maximum projections of apotome (Zeiss) optical sections captured through a depth of 3.5 µm of tissue. Survival = 5 weeks post injury/transplantation. All scale bars = 20 µm.
(A, B) High power images showing hMito+ hGDAsBMP in the process of migrating and accumulating at the pial surface within lateral funniculus white matter. Note the elongated radially orientated processes displayed by some hMito+ hGDAsBMP within white matter (B: arrowheads), a glial morphology indicative of tangential migration of these cells towards the adjacent pial surface (hMito: red; NF+ axons: green). (C) hGDAsBMP displaying typical astrocytic “stellate” arrangements of their processes within white matter immediately ventral to the injury site. Survival = 5 weeks post injury/transplantation. Scale bars: A, B = 100 µm; C = 20 µm.
Despite the similar ability of transplanted hGDAsBMP and
hGDAsCNTF to span the rostral to caudal extent of injury sites
and migrate into adjacent tissues, only hGDAsBMP promoted locomotor
recovery following transplantation into the transected dorsolateral funiculus
(DLF). This injury severs descending, supraspinal axons and causes chronic
deficits in both fore- and hind-limb motor function
Graphs show the average number of mistakes per experimental group made during Grid walk testing of locomotor recovery at 1 day before injury to 28 days after injury. In two separate experiments, hGDABMP transplanted animals (closed circles) performed significantly better than hGDACNTF (A) or hGPC (B) transplanted animals at all time points from 7 to 28 days post injury/transplantation. Note that the performance of hGDACNTF or hGPC transplanted animals was not significantly different from control injured rats at all time points (two-way repeated measures ANOVA, *p< 0.05).
We next examined the question of whether the precursor cells from which
hGDAsBMP were derived were also capable of promoting behavioral
recovery after DLF transection and found that pre-differentiation of these
precursor cells into astrocytes was essential to promote significant functional
recovery. Rats that received hGDABMP transplants performed
significantly better on the grid-walk test than either the hGPC transplanted
group or the media-injected control injury group at all time points from 7 to 28
days post injury/transplantation (
Differences in behavioral recovery were mirrored by marked differences in
promotion of neuronal survival within ipsilateral gray matter immediately
adjacent to sites of injury (
Montaged images of NeuN immuno-histochemistry at the C5 spinal level of
normal (A) and untreated injured (control) spinal cords (B) show that
the unilateral DLF transection injury causes loss of NeuN+ neurons
in multiple spinal cord laminae adjacent to the transected white matter.
Transplantation of hGDAsBMP (C) promotes significant
protection of neurons in laminae 7, 8, and 9 at the injury center. In
contrast, transplantation of hGDAsCNTF (D) or hGPCs (E) did
not promote significant levels of neuroprotection. (F) Schematic showing
gray matter laminae at the C5 level of the rat spinal cord (adapted from
(A) hGDABMP transplantation led to significant increases in numbers of NeuN+ neurons counted in a 1.8mm length of spinal cord encompassing the injury site. Graphs show percentage changes in numbers of NeuN+ neurons in laminae 4 to 9; laminae 4, 5, and 6; 7; and 8 and 9 in spinal cords from animals that received transplants of 9W2 or 9W1 hGDAsBMP, hGDAsCNTF or hGPCs and untreated control injuries. (B) Analysis of neuron survival within laminae immediately adjacent to the site of injury shows that hGDABMP transplantation promoted significant protection of neurons when all laminae were considered (4 to 9), with the most robust increases in neuron numbers in intermediate (7) and ventral (8 and 9) gray matter laminae. Numbers of NeuN+ neurons in spinal cords of rats transplanted with GDAsCNTF or hGPCs were not significantly different from each other or untreated spinal cord injuries. * ANOVA and Pairwise Multiple Comparison (Holm-Sidak), p<0.05. + t-test comparison with untreated spinal cord injuries, p<0.05.
Transplant | Lam. 4 to 9 |
Lam. 4 to 9 |
No Transplant | 2374.25+/−352.84 | 909.0+/−237.02 |
9W2 hGDABMP | 3313.75+/−238.51* | 1204.25+/−138.88* |
9W2 hGDACNTF | 2548.5+/−175.44 | 906.0+/−103.94* |
9W1 hGDABMP | 3128.25+/−327.41* | 1153.75+/−127.95* |
9W1 hGPC | 2578.2+/−71.59 | 938.6+/−117.48 |
ANOVA | p<0.001 | p<0.05 |
Lam. 4, 5, 6 | Lam. 7 | Lam. 8, 9 | ||||
|
|
|
|
|
|
|
No Transplant | 1191.25+/−259.127 | 456.50+/−124.66 | 948.75+/−119.29 | 330.5+/−74.25 | 359.25+/−108.53 | 122.0+/−45.35 |
9W2 hGDABMP | 1530.25+/−139.1 |
550.0+/−103.91 | 1242+/−95.62 |
447.5+/−41.99 |
541.5+/−61.47 |
206.75+/−22.40 |
9W2 hGDACNTF | 1178.75+/−126.64 | 396.0+/− 71.58 | 909.75+/−48.92 | 356.75+/−12.34 | 435+/−74.62 | 153.25+/−42.52 |
9W1 hGDABMP | 1466+/−151.15 |
528.50+/−59.04 | 1171+/−110.22 |
437.5+/−49.94 |
491.25+/−99.12 |
187.75+/−43.87 |
9W1 hGPC | 1248+/−35.02 | 435.2+/−61.72 | 870.4+/−74.65 | 336.6+/−50.97 | 459.8+/−70.28 | 166.8+/−49.43 |
ANOVA | p<0.05 | p = 0.09 | p<0.001 | p<0.05 | p = 0.08 | p<0.05 |
# All sections sampled above, at and below injury site.
* Significantly different from SCI animals receiving no transplant, Holm-Sidak post hoc test.
† Separate t-tests demonstrate that 9W1 and 9W2 GDAsBMP are statistically different from control injury (p<0.05).
Significant improvements in neuronal survival in all laminae studied were seen in
a 1.8 mm length of spinal cord encompassing the injury site of
hGDABMP-treated animals. Combined neuron counts for all laminae
(4 to 9) showed that hGDABMP transplantation promoted increases of
40% and 32%, in two separate experiments, of surviving neurons
compared to untreated injured spinal cords (
The present studies provide multiple novel findings relevant to the development of
astrocyte transplantation therapies for treatment of the injured or diseased central
nervous system. We show that subpopulations of human astrocytes, generated by
activation of different signaling pathways in the same population of human glial
precursor cells, have markedly different effects when transplanted into the injured
spinal cord. hGDAsBMP provided extensive benefit, including robust
protection of spinal cord neurons, increased support of axon growth and locomotor
recovery. In contrast, transplantation of either undifferentiated hGPCs or
hGDAsCNTF failed to provide significant benefits. The major gains in
behavioral recovery and neuronal survival achieved by
The development of astrocyte transplantation represents a new avenue for the
treatment of CNS injury, as contrasted with the extensive research that has been
conducted on replacement of oligodendrocytes. Starting with transplants of human
oligodendrocytes in the late 1980s
One of the striking differences in outcome between our studies and work on
oligodendrocyte and oligodendrocyte-precursor replacement lies in the finding that
differentiation of precursor cells into a specific astrocyte subtype
Along with demonstrating the marked benefits from astrocyte transplantation in
experimental injuries of the spinal cord, our studies also demonstrate that
obtaining benefit may require transplanting very specific populations of human
astrocytes. The significant difference in outcome achieved by transplantation of
hGDAsBMP versus hGDAsCNTF demonstrates clearly that not
all astrocytes are equivalent in respect to their therapeutic value, and this
appears to be the first study demonstrating functional differences between different
human astrocyte populations with respect to repairing the adult central nervous
system. It is also interesting to note the similarity between the outcomes obtained
with human cells and with our prior studies on rat cells
It was also of interest to observe that prolonged survival of the grafted astrocytes
was not required to obtain durable improvements in behavior and neuronal survival.
This also demonstrates a conservation of outcomes between human cells and rat cells,
which also did not require prolonged survival to provide durable benefit
This is also the first study, to our knowledge, in which transplanted astrocytes
(rodent or human) have been shown to promote extensive neuroprotection of spinal
cord neurons following spinal cord injury, an observation consistent with the robust
neuroprotective effects of intra-spinal rodent GDABMP transplants on
axotomized neurons of the red nucleus
The underlying mechanisms accounting for why hGDAsBMP are so much more
beneficial in terms of neuroprotection and functional recovery than either
hGDAsCNTF or undifferentiated precursor cells when transplanted into
spinal cord injured rats remain to be investigated, but it is likely that multiple
cellular functions are involved. For example, hGDAsBMP express higher
levels of such astrocyte-related genes as glutamate transporter 1, connexin 43, and
AKAP12, which are relevant to maintaining tissue homeostasis in the CNS as well as
reducing astrogliosis and neuronal death after injury, mediating glutamate uptake
and promoting blood-brain barrier formation
In brief our present studies provide the first demonstration of the utility of human astrocyte transplantation as a therapy for central nervous system injuries. Moreover, our studies provide a specific population of human astrocytes that appear to be particularly suitable for further development towards clinical applications.
The University of Rochester RSRB has reviewed this study and determined that based on federal (45 CFR 46.102) and University criteria, the study does not qualify as human subjects research and has waived the need for consent (RSRB#00024759). All animal procedures were performed under guidelines of the National Institutes of Health and approved by the Institutional Animal Care and Utilization Committee (IACUC) of University of Colorado Denver, Aurora, CO (UCAR# 80710(05)1E). or the IACUC of University of Rochester Medical Center, Rochester, NY (UCAR# 2008-075).
Human spinal cord tissues were obtained from two nine week old, de-identified
abortus samples collected in the course of medically prescribed procedures using
the Safe-Harbor Method. Spinal cord-derived glial precursors were grown and
isolated as previously described
Characterization of hGPC and hGDA cultures was performed by reverse-transcriptase
semi-quantitative polymerase chain reaction (RT-QPCR), Western blot and
immunofluorescent labeling as previously described
Adult female Sprague Dawley rats (3 months old, Harlan) were used in all
A total of 6 µl of hGDABMP, hGDACNTF or hGPC
suspensions (30,000 cells/µl; 180,000 cells total) per animal were acutely
transplanted into six different sites at the injury site on the right side of
the spinal cord: medial and lateral of the rostral and caudal injury margins,
and medial and lateral of the injury center (Supplemental
At 5 weeks post-surgery animals were deeply anesthetized and transcardially
perfused with 0.1 M PBS followed by 4% paraformaldehyde in 0.1M PBS.
Dissected spinal cords were cryosectioned and immunofluorescently labeled as
previously described
The relative density of axons within the centers of hGDABMP or
hGDACNTF transplanted injury sites was determined by quantifying
neurofilament-immunoreactive pixels in 4 tissue sections per spinal cord from 5
animals per experimental group. Images were captured (Zeiss-Z1 microscope) of
the right-side dorsolateral funiculus from every sixth histological cross
section (4 sections in total per spinal cord) from tissue at injury centers.
Using Image-J analysis software, a 465 µm×465 µm square region
of interest was drawn on each image with the upper right corner located on the
dorso-lateral outer edge of the transplant mass such that the region of interest
was contained within injury sites/transplant parenchyma (Supplemental
Neuronal survival within spinal cord gray matter was determined using NeuN
immuno-reactivity as a marker of surviving spinal cord neurons after spinal cord
injury
Behavioral analysis of volitional foot placement was tested using a grid-walk
behavioral test (Foot Misplacement Apparatus, Columbus Instruments) as
previously described
Schematic illustrations of the adult rat dorso-lateral funiculus (DLF)
transection spinal cord injury model and cell injections at injury sites.
Dorsal (A) and cross section (B) schematics of the rat cervical spinal cord
showing right side unilateral transection injury (red shaded area) conducted
at the level of the C3/C4 intervertebral junction. (C) Injections (six in
total) of either hGDAs or hGPCs were made at sites if injury, two into
injury centers and two further injections each to rostral and caudal injury
margins respectively (black diamonds represent injection sites). C3/C4,
junction of the third and fourth cervical vertebrae; DLF, dorsolateral
funiculus; Cf, cuneate fasciculus; Gf, gracile fasciculus; GM, gray matter.
(Cross section schematic (B) adapted from Grant and Koerber
(TIF)
Schematic illustration of neurofilament sampling region at injury/transplantation sites. Image-J analysis software has been used to draw a 465 µm×465 µm square region of interest (ROI, white box) on a representative image of an NF immuno-stained tissue section at the center of hGDACNTF treated DLF injury site. The upper right corner of the ROI is located on the dorso-lateral outer edge of the transplant mass such that that the region of interest is contained within the injury site/transplant mass. Scale bar = 200 µm.
(TIF)
We thank K. Saul, D. Harlow and K. Ellison for conducting histology and behavioral testing (Department of Neurosurgery, UC Denver) and Michelle Lacagnina (Department for Biomedical Genetics, University of Rochester, NY) and Laurie Baxter (Surgical Pathology, University of Rochester, NY) for assistance with human cell cultures, and Brendan Carlin for assistance with RT-PCR analysis (Department for Biomedical Genetics, University of Rochester, NY). The 3F8 anti-phosphacan antibody was obtained from the Developmental Hybridoma Bank developed under the auspices of the NICHID and maintained by the University of Iowa, Department of Biological Sciences.