Use of polysialic acid in repair of the central nervous system
- Laboratory of Cellular and Developmental Neuroscience, Department of Cell Biology, Memorial Sloan–Kettering Cancer Center, 1275 York Avenue, New York, NY 10021
-
Communicated by Lynn T. Landmesser, Case Western Reserve University, Cleveland, OH, September 19, 2006 (received for review February 17, 2006)
Abstract
Polysialic acid (PSA), a large cell-surface carbohydrate that regulates cell interactions, is used during vertebrate development to promote precursor cell migration and axon path-finding. The induction of PSA expression in damaged adult CNS tissues could help them to rebuild by creating conditions permissive for architectural remodeling. This possibility has been explored in two contexts, the regeneration of axons and the recruitment of endogenous neural precursors to a lesion. Glial scars that form at CNS injury sites block axon regeneration. It has been found that transfection of scar astrocytes by a viral vector encoding polysialyltransferase leads to sustained expression of high levels of PSA. With this treatment, a substantial portion of severed corticospinal tract axon processes were able to grow through a spinal injury site. In the studies of precursor cell migration to a cortical lesion, it was found that induced PSA expression in a path extending from the subventricular zone to a lesion near the cortical surface increased recruitment of BrdU/nestin-positive cells along the path and into the injury site. These displaced precursors were able to differentiate in a regionally appropriate manner. These findings suggest that induced PSA expression can be used as a strategy for promoting tissue repair involving both replacement of cells and rebuilding of neural connections.
Polysialic acid (PSA) is a long, linear homopolymer that is attached to the neural cell adhesion molecule (NCAM). It has the remarkable ability to down-regulate a wide variety of contact-dependent cell interactions. This activity is believed to stem from the ability of the charged polymer to occupy a large hydrated volume and thereby cause a direct physical hindrance of cell–cell contact (1, 2) (see Fig. 4, which is published as supporting information on the PNAS web site). The regulated expression of PSA has been shown to promote a reduction in cell interactions that can create permissive conditions for changes in the structure of a variety of tissues (see ref. 3 for review). That is, PSA does not in of itself specify the nature of the change that occurs, but rather it allows other factors to override existing cell interactions such as adhesion.
The broadest expression of PSA occurs on precursor cells during early development and plays a role in promotion of the migration of these cells after cell division (4–9). In most cases, PSA expression is lost after the migration process is completed. However, the axons of neurons often retain PSA to bundle, sprout, and branch appropriately during axon path-finding (10–17). Although most axons lose PSA upon forming synaptic contacts, within certain regions of the CNS, PSA can be retained by axons that are capable of synaptic remodeling (refs. 18–23 and see ref. 24 for review).
Results
Use of PSA as a Tool to Enhance Tissue Repair.
Given the nature of PSA's function during development, the potential exists that the induction of PSA expression in damaged adult tissues could help them to rebuild by creating conditions permissive for architectural remodeling. As a proof-of-principle, this potential has been explored in two contexts, the regeneration of CNS axons and the recruitment of endogenous neural precursors to a cortical lesion.
Regeneration of CNS Axons.
The failure of axons of the adult mammalian CNS to regenerate after lesion does not represent an intrinsic inability of CNS axons to grow but rather the nonpermissive nature of the CNS environment (25–29). For a brief time, the CNS is able to support sprouting of axons at the lesion site, but the growth cones soon adopt a swollen dystrophic morphology typical of growth inhibition (25, 27).
Interestingly, the most persistent axonal growth occurs in close apposition with a small subpopulation of astrocytes that transiently express PSA-NCAM after lesion (30–32). Expression of PSA-NCAM is known to block signals transmitted to neuronal cells (33), alter glial cell-based tissue architecture (34), and modulate growth cone behavior (17, 35). On this basis, we designed experiments to determine whether a sustained overexpression of PSA by astrocytes at the lesion site could interfere with the negative interaction between axons and the lesion site and thereby allow a continuation of regeneration beyond the abortive sprouting phase.
The right corticospinal tract (CST) of anesthetized 4- to 5-month-old male mice was lesioned completely by using the tip of a needle near the level of the 10th thoracic vertebra. To ensure complete severance of the right CST, the lesion was performed twice and extended vertically and laterally beyond the tract limits (Fig. 1 C–F). Subsequent cresyl violet staining confirmed that the lesion extended from the dorsal surface to the central canal and included the entire CST as well as ascending sensory pathways (Fig. 1 E).
Design of axon regeneration studies. (A) Constructs. (Upper) Map of the lentiviral TVA-specific transfer vector carrying GFP (control). (Lower) Viral construct carrying the GFP-PST fusion protein. (B) Overall approach. (Upper) Drawing showing corticospinal axons (blue) projecting from the cerebral cortex through the spinal cord. A complete section of the right CST was performed near the 10th vertebra, and the viral suspension was locally injected at the time of lesion. (Lower) The distal part of axons degenerated and a nonpermissive glial scar formed at the lesion site. The CST was labeled red by cortical injection of fluororuby. (C) Transverse section of the lesion. The blue line delineates the lesion extent in a spinal cross-section. The entire right CST (red) was severed as well as the ascending sensory pathways, the left CST, and parts of the dorsal horns. The single-headed orange arrow indicates the insertion path of the needle. The double-headed orange arrow indicates the movement of the needle used to lesion the CST. This procedure was carried out twice to ensure complete transection of the CST. CC, central canal. (D) Lesion site. Dorsal view of spinal cord showing the location of the right unilateral lesion of the dorsal white matter as indicated by its brownish color. The red lines indicate the different levels of axonal quantification (1.5 mm proximal to the lesion, and 0.5, 1.5, and 2 mm distal to injury). (E) A cresyl violet-stained sagittal section of the spinal cord showing the lesioned area. The lesion extended from the dorsal surface to the central canal (CC) and completely interrupted the CST and the ascending pathways (AP). Boundaries of the lesion are taken as its most rostral and caudal margins. (F) Viral infection of lesion astrocytes. Cross-section at the infected lesion site showing GFP expression by infected astrocytes. Note that the lesion (circled area representing the extent of gross tissue damage) encompassed the entire right CST and most of the right dorsal horn as well as a large part of the left dorsal white and gray matter. The dense bright areas are densely packed infected astrocytes within the scar. The lesion went as deep as the central canal (CC). The ventral horns are delineated and indicated by asterisks. (Scale bar: 50 μm.) (G) GFAP immunostaining in the lesion area. GFP expression (green) occurred exclusively in areas that were immunopositive for GFAP (blue), which attests to the specificity of the virus for astrocytes.
To achieve a selective and sustained overexpression of PSA on astrocytes at the lesion site, an astrocyte-specific HIV(ALSV-A) virus vector (36, 37) carrying the PSA-synthesizing polysialyltransferase (PST) ST8Sia IV was injected at the lesion site shortly after the injury (Fig. 1 A and B; see Fig. 5, which is published as supporting information on the PNAS web site). Such a PST gene introduction is known to produce physiological levels of PSA specifically on NCAM (34). GFP virus was used as a control. Six weeks later, a large number of GFP-expressing astrocytes were found at the lesion site, and infection was observed to be restricted to astrocytes as shown by exclusive localization of GFP in glial fibrillary acidic protein (GFAP)-stained cells (Fig. 1 F and G). Immunostaining for PSA revealed that infection with the GFP-PST virus resulted in high levels of PSA on the cell body and processes of astrocytes in the scar region (Fig. 2 A).
PST induces PSA overexpression on astrocytes and renders the tissue permissive for axon regeneration. (A Left) Images from control samples injected with GFP virus. Note the green GFP-expressing astrocytes at the lesion site. (A Right) Scar tissue from GFP-PST-treated animals showing high levels of PSA (red). (Scale bar: 20 μm.) (B) Image from a control sagittal section showing fluororuby-labeled CST axons (red) at proximal edge of the lesion. Only a few sprouts were able to enter the scar. GFP-expressing astrocytes are green. (C) Images of CST axons (red) growing in contact with cells expressing both GFP (green; Upper) and PSA (blue; Lower) within the scar of a GFP-PST-infected spinal cord. The arrow indicates the rostro-caudal axis. (Scale bar: 20 μm.) (D Left) Representative images of swollen dystrophic growth cones that were found at the entrance of the lesion site in control GFP-infected animals. (D Right) Examples of growth cones in GFP-PST-infected spinal cords. Of 380 nondystrophic growth cones observed in the GFP-PST infected animals, ≈90% were located in PSA-rich areas. In PSA-poor regions of the same animals, most of the growth cones were swollen. (Scale bar: 10 μm.) (E and F) Representative three-dimensional, horizontally rotated reconstructions of 100-μm-thick transverse sections of the CST located 0.4–0.5 mm distal from the lesion. Note the numerous axon profiles in GFP-PST-treated samples as compared with GFP-treated controls. (G) Quantification of axon regeneration in spinal cross-sections distal to the lesion. A significant number of axons (representing on average 10% of the labeled axons proximal to the injury) were able to cross the lesion gap and grow more than 0.5 mm beyond the lesion in GFP-PST samples. Note that after the axons entered the distal segment of the spinal cord, they were inhibited from growing to 1.5 mm beyond the scar.
To observe regeneration, CST axons were anterogradely labeled 5 weeks after lesion by stereotaxic injection of fluororuby in the contralateral sensory-motor cortex; ≈1,000 labeled axons (Table 1) were found at the entrance of the lesion site. The animals were killed 6 weeks after surgery by transcardial perfusion with paraformaldehyde, and their spinal cords were processed for histological evaluation in either transverse sections for quantitation or sagittal sections for histological characterization.
Quantification of individual fluororuby-labeled axons in the CST proximal and distal to the lesion in five control (GFP) and five experimental (GFP-PST) animals
With control GFP virus, labeled processes rarely entered the lesion site and exhibited swollen dystrophic growth cones (Fig. 2 B). In contrast, with GFP-PST virus, labeled axonal processes readily were observed growing across the PSA-rich tissue (Fig. 2 C). Within the GFP-PST virus-infected animals, there were regions with higher and lower levels of PSA expression, and nearly all of the nondystrophic growth cones were found in PSA-rich regions (Fig. 2 D and Fig. 6, which is published as supporting information on the PNAS web site). Quantification of fluororuby-labeled CST axonal processes by confocal microscopy indicated that the number of labeled processes that were found 0.5 mm beyond the most distal extension of the lesion site was 6–15% of the number that were found 1.5 mm proximal to the margin of the injury. In GFP virus-infected animals, only an occasional process (three to four per lesion site) exhibited this capability (Table 1; Fig. 2 E and F). Estimates of labeled fibers midway through the lesion were 474 ± 92 in PST-treated samples and 56 ± 30 in controls. Once the labeled axons had grown beyond the PSA-rich scar, they encountered a new environment that includes degenerated axons and their myelin. These axons were unable to travel more than 1.5 mm beyond the scar boundary (Fig. 2 G), suggesting that they were being affected by inhibitory properties in this environment (ref. 38 and see ref. 28 for a recent review).
Although from these results it is clear that axon behavior in the scar has been substantially altered by PSA expression, it remains to be determined to what degree the quantitation obtained reflects regeneration of individual axons as opposed to growth of multiple sprouts derived from a smaller number of axons. The cellular mechanism by which the PSA-expressing scar tissue has been rendered permissive for process growth also remains to be determined. Given the heterogeneity of scar tissue, it would be difficult to detect subtle changes in the apposition of scar astrocytes that might allow axon penetration. The possibility also exists that the action of PSA occurs at the level of contact-dependent cell signaling. That is, PSA could buffer the detrimental effects of growth inhibitory molecules present in the scar tissue, such as ephrins (39) or chondroitin sulfate proteoglycan (40, 41). Interestingly, although chondroitin sulfate proteoglycan and PSA are both large, negatively charged glycans, they appear to have the opposite effect on axon regeneration and thus must operate via distinct mechanisms; we also observed that PSA expression does not alter chondroitin sulfate proteoglycan expression after injury (Fig. 7, which is published as supporting information on the PNAS web site).
Recruitment of Endogenous Neural Progenitors to a Brain Injury.
Neural progenitors are continuously born in the adult mammalian brain, where they migrate to replenish particular cell populations, such as the interneurons of the olfactory bulb and granule cells of the dentate gyrus (42). Precursors also can populate regions of the adult neocortex and striatum after injury in rodents (43–45). However, the number of new cells that reach injury sites is small and unlikely to allow adequate restoration of function. Given the ability of PSA to facilitate precursor cell migration, it was of interest to determine whether PST-induced expression of PSA also could facilitate recruitment of precursors from the subventricular zone (SVZ) to lesion sites. For this purpose, the PST gene delivery system described above was used to produce PSA on astrocytes along a needle-track pathway extending from the SVZ to the cortical surface, passing through a focal apoptotic lesion produced by injection of ibotenic acid (Fig. 3 A and C). High levels of PSA were obtained on astrocytes all along the track (Fig. 3 D and E), whereas the chemical lesion alone induced only a small amount of PSA in control samples.
PSA overexpression promotes progenitor recruitment into the chemically induced cortical lesion. (A) Light microscope image of a sagittal brain section. The line shows the injection path. The red area indicates the SVZ and the adjacent rostral migratory stream. (Magnification: ×10.) (B) Diagram showing the injection of ibotenic acid in the cerebral cortex and the needle track (arrow) with virus infecting a region reaching from the SVZ to the cortical lesion. CC, corpus callosum. (C) Annexin-V immunostaining indicates that injection of ibotenic acid in the cerebral cortex produced tissue apoptosis that remained detectable 3 weeks after injection, whereas injection of PBS induced no apoptosis. (Inset) A higher magnification of the apoptotic area in the box. (Scale bar: 200 μm; Inset: 20 μm.) The white line marks the cortical surface. (D) Injection of PBS alone did not induce PSA expression, whereas low levels of PSA could be detected after a combination of ibotenic acid and injection of the control virus. As expected, addition of the PST virus produced high levels of PSA expression. (Scale bar: 200 μm.) The white line indicates the cortical surface. The parallel lines indicate the injection route. (E) Sample injected with the PST virus. A complete correlation between GFAP and PSA expression indicates that GFAP-positive cells were the only cells expressing PSA 30 days after injection. (Scale bar: 20 μm.) (F) The number of BrdU/nestin-positive progenitors found in the ibotenic acid-injected cerebral cortex was markedly higher in PST-infected samples as compared with controls. Note that the recruited progenitors were found more frequently along the needle track but also were dispersed tangentially over the apoptotic area. (Scale bar: 100 μm.) CC, corpus callosum; m, meninges. The thin parallel white lines indicate the injection track. (G) (i) Sagittal section of an adult mouse brain. The white line shows the injection route, and the box crossing the line indicates the center of the lesion. (ii Left) NF-M/BrdU double-stained cells could be observed in the lesion lateral to the injection route. (ii Right) NF-M/BrdU-labeled cells at the epicenter of the injection. (iii) Higher magnification of the Insets in ii. (Scales bar in ii: 20 μm; in iii: 10 μm.) (H) Percentage of BrdU-positive progenitors that migrated into the lesioned cortex and coexpressed NF-M (neurons), CNPase (oligodendrocytes), or GFAP (astroglia). Most progenitors differentiated into more mature NF-M-positive neurons. Only cells that were positive for both BrdU and the second marker in the orthogonal projection were counted (n = 8; total BrdU-positive cells = 1,331). (I) Quantification of the number of BrdU/nestin-positive cells in the chemical cortical lesion showing that the lesion alone (control) recruited a small but significant number of progenitors as compared with uninjected cortices. Infection of the needle track astrocytes with the PST virus resulted in an additional large increase in progenitor recruitment into the cortical lesion (approximately six times, P < 0.001). Values are the mean ± SEM of migrating cells counted in five to seven 40-μm brain slices. n = 8 animals. (J) BrdU-positive cells in the corpus callosum (CC) of animals treated with PST also were immunopositive for the oligodendrocytic markers O4 (early oligodendrocytes) and CNPase (mature oligodendrocytes). (Scale bar: 50 μm.) The arrows indicate the process of a CNPase-positive cell. (Scale bar: 20 μm.) (K) Quantitation of the percentage of BrdU-positive cells coexpressing CNPase and/or O4 (oligodendrocytic markers), GFAP (astrocytes), NeuN (early postmitotic neurons), or NF-M (mature neurons) in the lesioned corpus callosum (CC) of animals injected with PST virus. Only cells showing nuclear BrdU staining and a cell-type marker in the orthogonal reconstructions were included in the quantification. Note that the sum of the percentages for CNPase and O4 exceeded 100%, suggesting their coexpression in some cells. The graph shows the percentage of total BrdU-positive cells that were double-labeled for each cell marker. Most progenitors differentiated into mature oligodendrocytes (expression of CNPase), and only very few were able to become mature neurons (expression of NF-M). n = 8; total BrdU-positive cells = 221.
Although the control injection protocol resulted in recruitment of only a small number of BrdU/nestin-positive cells along the path from the SVZ to the cortical lesion, injection of the PST construct increased these numbers by more than six times (P < 0.001) (Fig. 3 F and I). There also was what appeared to be a robust tangential dispersion of these cells corresponding to the region occupied by chemical lesion and PSA expression (Fig. 3 F). Neither the needle track nor the chemical lesion appeared to affect the normal migration pattern of SVZ cells (data not shown). Interestingly, if the PSA induction protocol was performed in the absence of the cortical lesion, the SVZ cell migration only extended as far as the neighboring corpus callosum (Fig. 8, which is published as supporting information on the PNAS web site). This finding suggests that the cortical lesion is exerting an attractive influence on the SVZ cells and that PSA's major role is to facilitate a migratory response to that attraction.
Recruitment of precursor cells into damaged tissue is useful only if they are able to differentiate into cell types appropriate to restoration of function. The differentiation fate of the BrdU-labeled precursor cells diverted from the SVZ was evaluated both in corpus callosum adjacent to the SVZ and in the cortical lesion, by immunostaining for cell-type-specific markers. In the corpus callosum, a substantial portion of the BrdU-labeled cells expressed oligodendrocyte markers CNPase and/or O4 (Fig. 3 J and K; Fig. 9, which is published as supporting information on the PNAS web site). Only 4% of BrdU-positive cells were GFAP-labeled, and only 1% of GFAP-expressing cells were BrdU-positive, indicating that the experimental manipulations had not led to a significant proliferation of glial cells (Fig. 10, which is published as supporting information on the PNAS web site). Although one-third of the cells expressed NeuN, they did not express NF-M (data not shown). Thus, most of the precursor cells diverted to the corpus callosum followed an oligodendrocytes lineage, with the remainder apparently delayed or unable to become mature neurons. In contrast, in the cortical lesion, 74% of the BrdU-positive cells were found to be NF-M-positive, whereas 23% were CNPase-positive and 4% were GFAP-positive (Figs. 3 G and H and 9). Therefore, most of these progenitors in cortex had followed a neuronal fate. Additional neuronal markers showed that 23% of BrdU-positive cells in the cortex expressed choline acetyltransferase, 7% expressed glutamate decarboxylase, and 2% were tyrosine hydroxylase immunopositive (data not shown).
Therapeutic Potential.
The fact that PSA overexpression on astrocytes promotes growth of corticospinal axon processes through the lesion site suggests that this procedure could serve as part of a therapy for recovery from injury-induced paralysis. That is, the induction of PSA at the site of damage could help axons to regrow beyond the injury, at which point the second major obstacle to regeneration, inhibition produced by the environment beyond the injury site, can be more effectively addressed. Thus, it is reasonable to anticipate that with a combination of PSA expression at the scar with strategies to reduce inhibition by the distal environment (29, 46), significant numbers of axons may be able to reach the vicinity of their targets. Additional evaluation of this potential will require a more precise assessment of the extent of multiple sprouting of individual axons, optimization of the effects obtained, extension to other injury models, and development of a more therapeutically compatible delivery system for PSA (see below).
Our initial studies also suggest that induction of PSA on glial cells on a route connecting a cortical lesion with the SVZ could have two major benefits with respect to repair therapies based on endogenous stem cells: it increases the number of progenitors available to the injury site and provides them in a state amenable to differentiation according to local environmental cues. These results provide a basis for further investigations, including application to specific CNS lesions such as stroke. Interestingly, it appeared that the precursor cells that arrived at the lesion site were able to spread throughout the PSA-positive damaged area. This raises the possibility that PSA expression could be used in combination with exogenous stem cell approaches (for example, to Parkinson or Huntington diseases) to improve the dispersal of the injected cells into host tissues.
An important consideration with any therapy is the potential for deleterious side effects. The mechanism of action of PSA is to reduce cell interactions, which then creates permissive conditions for change. In fact, without additional factors to fuel that change (such as a stimulus for axon growth or cell migration), we have observed that the presence or absence of PSA by itself has little effect on tissue structure. Therefore, although each type of application will need to be evaluated for side effects, it is reasonable to expect that they would be acceptable relative to the possible benefits. It also should be noted that PSA expression should only need to be maintained during the repair process.
The present studies, which are offered here as an initial proof-of-principle rather than a demonstrable therapy, use a gene delivery system that is obviously not relevant to the clinic. In moving closer to application, it will be necessary to develop a different protocol to induce PSA, for example a PST-lentivirus vector that infects a broader range of cells and species. Other strategies that do not involve gene delivery also are worth exploring, including direct injection of a PST with its nucleotide sugar substrate (CMP-sialic acid), constructing a soluble protein–PSA conjugate that binds to the cell surface, or screening for small molecules that can stimulate PSA synthesis.
Materials and Methods
See Supporting Detailed Methods, which is published as supporting information on the PNAS web site, for further details.
Axon Regeneration.
Surgery.
The right CST of adult GFAP-TVA mice (37, 47) was completely transected near the level of the 10th thoracic vertebra. To selectively express PSA in TVA-expressing astrocytes, a TVA-specific HIV(ALSV-A) virus (36), carrying GFP-PST or GFP, was injected locally at the time of lesion.
Quantification of regeneration.
CST axons were anterogradely labeled 5 weeks after lesion by cortical injection of fluororuby (Molecular Probes, Eugene, OR). The animals were perfused with 4% paraformaldehyde at 6 weeks, and 100-μm spinal transverse sections were cut. Labeled axons located in the right CST were counted 1.5 mm proximal to the lesion. Counts inside the right CST also were performed at 0.5, 1.5, and 2 mm distal to the lesion site. To visualize growth cones, other samples were sectioned sagittally.
Immunostaining.
Immunodetection of PSA (5A5; 1:2,000) and GFAP (5 μg/ml; Chemicon International, Temecula, CA) was performed on floating sections (17, 33). After confocal imaging, the sections were restained with cresyl violet.
Recruitment of Endogenous Progenitors to Lesions.
Cortical injection of virus and ibotenic acid.
A needle was introduced 3.8-mm-deep through the cortex and corpus callosum to the edge of the SVZ of adult GFAP-TVA-mice. A GFP-PST or GFP HIV(ALSV-A) viral solution was injected during retraction of the needle toward the surface. Ibotenic acid then was injected at a depth of 1.5–2 mm to induce a local chemical lesion. The animals were perfused with 4% paraformaldehyde 30 days later.
Counting of migrating precursor cells.
Progenitors were visualized by BrdU labeling and nestin immunostaining then counted in consecutive 40-μm sagittal sections.
Cell markers.
Immunostaining for nestin (0.5 μg/ml), GFAP (5 μg/ml), NeuN (1:150), CNPase (1:150), NF-M (1:200), and O4 (1:150) was performed on 40-μm sagittal sections. All markers were obtained from Chemicon International (Temecula, CA) except for CNPase, which was obtained from Sternberg Monoclonals (Lutherville, MD).
Acknowledgments
We thank Dr. Brian Lewis (Memorial Sloan–Kettering Cancer Center), Dr. Eric Holland (Memorial Sloan–Kettering Cancer Center), and Dr. Harold Varmus (Memorial Sloan–Kettering Cancer Center and National Cancer Institute) for kindly providing the plasmids and transgenic animals.
Footnotes
- *To whom correspondence may be addressed. E-mail: u-rutishauser{at}mskcc.org or a-el-maarouf{at}ski.mskcc.org
-
Author contributions: A.E.M. and A.K.P. contributed equally to this work. A.E.M., A.K.P., and U.R. designed research; A.E.M. and A.K.P. performed research; U.R. contributed new reagents/analytic tools; A.E.M., A.K.P., and U.R. analyzed data; and A.E.M. and U.R. wrote the paper.
-
The authors declare no conflict of interest.
- Abbreviations:
- PSA,
- polysialic acid;
- NCAM,
- neural cell adhesion molecule;
- CST,
- corticospinal tract;
- GFAP,
- glial fibrillary acidic protein;
- PST,
- polysialyltransferase;
- SVZ,
- subventricular zone.
- © 2006 by The National Academy of Sciences of the USA