Apart from tailed amphibians and fish, which retain remarkable regenerative capabilities throughout life (Zottoli et al., 1994; Clarke and Ferretti, 1998; Ferretti et al., 2003), adult vertebrates are incapable of functional repair of the spinal cord after injury. During development, however, the spinal cord of chicks and opossums is able to regenerate (Shimizu et al., 1990; Hasan et al., 1991; Nicholls and Saunders, 1996). It is likely, therefore, that all vertebrate systems initially possess the ability to effectively repair the injured spinal cord and that subsequent developmental changes lead to the loss of this capability before birth.
Chick embryos can regenerate their spinal cord until approximately embryonic day (E) 13. The ability of the chick spinal cord to regenerate before E13 cannot simply be attributed to the overall immaturity of the nervous system. Neurogenesis in the spinal cord is virtually complete by E5 (Hollyday and Hamburger, 1977), and descending axonal projections from the brain are complete by E10 (Okado and Oppenheim, 1985). Oligodendrocytes, as detected by O4 immunostaining, are first evident around E5–E6, but the onset of myelination occurs several days later, around E12 (Ono et al., 1995). It has been demonstrated by a series of retrograde labeling experiments that true regeneration does occur after spinal transection in the embryonic chick (Hasan et al., 1993). Failure to regenerate is at least in part due to the generation of a nonpermissive environment within the adult central nervous system (CNS). Delaying myelination, for example, results in effective repair of the spinal cord much later in development, implicating proteins associated with the myelination process in the creation of a nonpermissive environment for successful axonal regrowth (Keirstead et al., 1995).
Many factors have been suggested to contribute to the inhibitory nature of the injured CNS. These factors include chondroitin sulphate proteoglycans (Zuo et al., 1998; Bradbury et al., 2002), myelin-associated glycoprotein (McKerracher et al., 1994; Mukhopadhyay et al., 1994), and Nogo (Varga et al., 1995). Neutralization of Nogo function with specific antibodies appears to enhance axonal extension in vitro (Spillmann et al., 1997) and in vivo (Brosamle et al., 2000).
Nogo belongs to the family of reticulon proteins, and three isoforms have been identified (Fig. 1), each of which shares a common C-terminus (Chen et al., 2000; GrandPre et al., 2000; Prinjha et al., 2000). Nogo contains two large transmembrane domains of around 30 amino acids, permitting several theoretical orientations across the cell membranes (Chen et al., 2000; GrandPre et al., 2002). Conflicting opinions exist regarding whether the N-terminal region of Nogo-A is located extra- or intracellularly (Huber and Schwab, 2000). The 66 amino acid domain (Nogo-66) present between the two transmembrane domains is believed to be located, at least partially, extracellularly, where it can mediate its inhibitory effects on axonal outgrowth by means of the Nogo receptor, NgR (Fournier et al., 2001). In addition to this active region shared by all Nogo isoforms, Nogo-A also possesses at least one other active domain located within the N-terminal region capable of exerting inhibitory effects on axonal outgrowth. This N-terminal domain also has effects on fibroblast spreading in vitro, although the receptor through which it operates remains elusive.
Of interest, the Nogo molecule also features a double lysine endoplasmic reticulum (ER) -retention motif, suggesting that, as for other reticulons, the primary location of Nogo may be within the ER membranes (Chen et al., 2000; GrandPre et al., 2000). The recent identification of a Nogo-interacting protein within the mitochondria also suggests a putative role for the Nogo family in regulation of cellular metabolism (Hu et al., 2002).
Expression of Nogo isoforms has been found to be largely restricted to oligodendrocytes in the adult, though a subset of neurons has been shown to express Nogo-A (Liu et al., 2002). A thorough developmental analysis, however, has yet to be reported. The aim of this study was to determine whether the onset and patterns of expression of Nogo-A and of NgR are consistent with the hypothesis that these molecules contribute to the loss of regenerative potential seen during development.
To this purpose, we have used the embryonic chick model, because the time when the spinal cord loses its regenerative ability has been well defined in this system, and the human embryonic spinal cord, both in vivo and in vitro, to study expression of Nogo-A and NgR by immunohistochemistry, Western blotting, and reverse transcriptase-polymerase chain reaction (RT-PCR). We have found that both Nogo-A and NgR are expressed in chick and human spinal cords at early stages of development during which, at least in the chick, functional regeneration can occur. Furthermore, analysis of spinal injuries performed in the chick during the permissive period for regeneration has indicated that Nogo-A is closely associated with the site of injury. Overall, these data suggest that Nogo and NgR play a role in the early development of the spinal cord and other tissues and that they do not inhibit axonal re-growth at developmental stages permissive for regeneration.
Isolation of Chick Nogo mRNA and Its Expression in Developing Spinal Cord
To address the question of whether the presence of Nogo contributes to the failure of spinal cord regeneration in adulthood and late stages of development, we examined its expression in a system in which the transitional point from regeneration permissive to restrictive states has been precisely determined. The embryonic chick is known to lose its potential to effectively repair the spinal cord around E13. Analysis of E11 spinal cord RNA by RT-PCR using primers designed from published chick EST sequences indicated the presence of Nogo-A in embryonic chick spinal cord (Fig. 2A). Further evidence for the existence of a chick Nogo-A was obtained by using additional primers to amplify and sequence a longer Nogo-A–specific region in the chick. Translation of the 1-kb fragment isolated revealed several highly conserved regions and an overall 45–47% amino acid identity with human (Fig. 2B) and rat (not shown) Nogo-A, but no obvious similarity with any other protein. A second 298-bp sequence was isolated using primers designed to amplify part of the C-terminal end of human Nogo (Fig. 2C). This region was found to be more highly conserved than the N-terminus and the highest percentage of amino acid identity (90%) was with Nogo-A rather than with mammalian Nogo proteins or other members of the reticulon family.
RT-PCR analysis of spinal cord samples using primers that amplified the N-terminus region demonstrated the presence of Nogo-A mRNA in dissected neural tube as early as E3 embryos, when only neurons have been born, and at all subsequent stages of spinal cord development tested (Fig. 3A).
Expression and Cellular Localization of Nogo-A at Early Stages of Chick Development
To determine the precise spatiotemporal distribution of Nogo-A protein in the developing chick spinal cord, immunohistochemistry was performed on transverse paraffin sections of embryos ranging from E3 to E18. The rabbit AS472 antibody used in this study was generated by using the rat Nogo-A–specific peptide sequence shown in Figure 2D (Chen et al., 2000). This sequence shares no homology with any other known gene sequence but is similar to the corresponding chick and human Nogo-A region (56% and 78% identity, respectively). Surprisingly, Nogo-A was detected in the most ventrolateral region of the E3 spinal cord, in the myotome and weakly in the notochord (Fig. 4A). Nogo-A expression extended dorsally within the spinal cord at E4 (not shown) and E5 (Fig. 5A), and the entire marginal layer was positive by E7 (Fig. 5B). Strong staining was observed in the developing white matter at E10 (Fig. 5C), E11 (Fig. 4C), and E13 (Fig. 5D), and was maintained at E15, albeit at lower levels (not shown). At E10, Nogo-A expression was also noted in peripheral nerves, in the enteric plexus (Fig. 5C) and arteries (not shown).
Because at certain developmental stages Nogo-A staining seemed localized on axons, we double-stained the developing spinal cord from E3, E11, and E17 embryos for Nogo-A and the high molecular weight neurofilament protein to confirm this observation (Fig. 4). These experiments clearly indicate that expression of the two proteins closely overlaps in the white matter at E3 and E11 (Fig. 4A–F). In contrast, at E17 no significant colocalization of Nogo-A and neurofilaments was observed within the spinal cord (Fig. 4G,H). This finding is consistent with previous studies showing that, in mature spinal cord, Nogo-A is expressed by oligodendrocytes (Chen et al., 2000; GrandPre et al., 2000).
Immunoblotting was performed to assess whether the presumptive chick Nogo-A protein is similar in size to Nogo-A in other species (Fig. 3B). Samples of E11, E13, E15, and E17 chick spinal cords were probed and a band of approximately 220 kDa, correlating well with the size of rat Nogo-A (Chen et al., 2000), was detected by the AS472 antibody. Another band of approximately 270 kDa was also observed. This finding is not totally surprising, as additional bands are detected by this antibody also in other species, although it seems to be a highly specific reagent for immunohistochemistry studies (Chen et al., 2000; GrandPre et al., 2000).
Expression of Nogo-A in Response to Spinal Cord Injury at Regeneration-Permissive Stages
Given that antibodies that block Nogo improve regenerative capability in normally nonregenerating spinal cords (Brosamle et al., 2000), we examined the possibility that Nogo-A may be down-regulated after injury to the spinal cord during the permissive period for regeneration. Analysis of longitudinal sections of chick spinal cord 24 hr after injury at E11 did not show any significant down-regulation of Nogo-A; strong staining was observed between the opposed stumps of the severed cord (Fig. 5M). Therefore, Nogo-A expression per se does not seem sufficient to inhibit regeneration.
Developmental Regulation of Nogo Receptor (NgR) in Chick Spinal Cord
Because Nogo can exert inhibitory actions through NgR, developmental regulation of this receptor could contribute to the transition from permissive to restrictive regenerative states. Therefore, we examined its expression in the developing chick spinal cord. Immunohistochemical analysis revealed little or no positive staining within the spinal cord at E3, although the notochord was clearly positive (not shown). Between E4 and E9, staining became more widespread in the differentiating gray matter and was also present in neurons within the dorsal root ganglia (Fig. 5E,F). Neuronal expression of NgR was not surprising, because in chick spinal cord, cultures sprouting axons are stained by this antibody (Fournier et al., 2001). That axonal staining was more obvious in cultured spinal cord than in sections might be due either to easier visualization of processes in vitro, or possibly to changes in the intracellular distribution of the receptor as a consequence of changes in cell–cell interactions in culture. Most spinal neurons were positive by E10 (Fig. 5G). NgR expression persisted in the neurons of the gray matter and DRGs until E13 (not shown). By E14 (Fig. 5H), however, and time points thereafter, NgR labeling was limited to a few individual neurons of the spinal cord. Analysis of the transcript by RT-PCR in dissected spinal cord confirmed presence of the transcript (Fig. 3C). A distinctive temporal pattern of expression was also noted in the heart and developing musculature. The heart was positively labeled by E3 (Fig. 5J) and the myotome by E5 (Fig. 5E). Skeletal muscle was first noted to express NgR at E9 (Fig. 5K), and by E10, all muscle fibers examined were positively labeled (Fig. 5G). The pattern of muscle staining mirrored that of neuronal labeling in the spinal cord, and by E14, no muscle fibers were expressing NgR (data not shown). Retinal expression was also observed, and as shown in sections of E5 chick embryo, it seemed to be restricted to the peripheral retina (Fig. 5I). In all cases, analysis of NgR staining under high magnification revealed a dotted staining pattern typical of membrane receptors (Fig. 5L). With the exception of early developmental stages where NgR expression was seen in the notochord (not shown) and myotome (Fig. 5E), by and large the localization of NgR appeared to be mutually exclusive with that of Nogo-A (e.g., compare Fig. 5A–D, with 5E–H).
Analysis of Nogo Localization in Developing Human Spinal Cord In Vivo and In Vitro
To assess whether early Nogo expression is a general feature of vertebrate development, RT-PCR analysis was performed on embryonic human tissue samples by using primers designed to specifically amplify Nogo-A and NgR (Fig. 6). In the earliest sample analyzed (45 days of gestation [dg]), Nogo-A cDNA was detected, but NgR was not present. The presence of both transcripts was confirmed in 10 (Fig. 6) and 13 (not shown) week human spinal cord samples. Myelin basic protein (MBP) expression was also examined to determine the relative maturity of oligodendrocytes within the spinal cord at the stages investigated. Only trace levels of MBP were found in the 45 dg spinal cord sample, with robust expression detected at 10 and 13 weeks gestation (not shown). The integrity of each cDNA sample was established by performing PCR with primers specific for the housekeeping gene β-actin.
To determine the precise localization of Nogo-A protein in the human embryonic spinal cord, we analyzed by immunohistochemistry transverse paraffin sections of embryos ranging from 28 to 70 days of gestation. Immunoblotting confirmed that the AS472 antibody used in these experiments was binding to a protein of the expected size for human Nogo-A (220 kDa) in spinal cord and muscle (Fig. 6B). No staining in the spinal cord was observed in 28-day embryos, although at this developmental stage, Nogo-A was present in the notochord, the myotome, and epidermis (Fig. 7A). By 45 days of gestation, cells of the ependymal canal stained positively for Nogo-A (Fig. 7B). The developing white and gray matter were first noted to express Nogo at 55 days of gestation and immunostaining appeared to be strongest in the ventral half of the spinal cord (not shown). At later stages (61–70 days of gestation), the levels of Nogo protein in the ventral spinal cord appeared to have increased and Nogo was expressed also in more dorsal regions of the cord and in the dorsal root ganglia (Fig. 7C,D). Strong immunostaining was also present in the striated musculature of the heart and trunk (Fig. 7K,L) and around the esophagus (not shown). A very similar pattern of reactivity was observed by using an antibody that recognizes both Nogo-A and Nogo-B (not shown), suggesting that Nogo-A is the predominant Nogo isoform expressed during development in humans. The possibility that Nogo-B is coexpressed with Nogo-A in most of the tissues examined, however, currently cannot be ruled out.
To further analyze which cell types predominately express Nogo and whether the N-terminus region is intracellular or extracellular, we performed immunocytochemistry on explant cultures of embryonic human spinal cord of various ages using antibodies raised against the N-terminus (see Fig. 1). When unfixed spinal explants were stained using anti–Nogo-A antibodies, very few cells were labeled (Fig. 8A,B). Double-staining procedures revealed that these positive cells did not express O4, glial fibrillary acidic protein (GFAP), or neurofilament proteins (data not shown). In contrast, after fixation and permeabilization of cultured spinal cords, many more cells, several of which had a broad, flat morphology were Nogo-positive (Fig. 8C,D). In permeabilized cells from embryos of 41–61 days of gestation, Nogo-A colocalized with neurofilament (Fig. 8E) but not with GFAP or O4 (Fig. 8F,G). In contrast, experiments performed on fixed cells from later stage spinal cords (98 days of gestation) resulted in double labeling of Nogo with O4 (Fig. 8H). Overall, these data indicate that Nogo is up-regulated in oligodendrocytes at later stages of development and that the N-terminal region of Nogo-A is chiefly located intracellularly, although no distinctions can be made between a cytoplasmic or ER location.
Localization of NgR in Developing Human Spinal Cord
We examined also the tissue distribution of NgR during human embryonic development by immunohistochemistry (Fig. 7E–J). The antibody detected a single band in Western blots of human embryo extracts (not shown). At 28 days of gestation, the receptor was weakly expressed in myotome and notochord (Fig. 7E) and highly expressed in the heart (Fig. 7J). By 45 days of gestation, expression was evident in the developing dorsal root ganglia, and in a few individual cells of the gray matter (Fig. 7F). In the spinal cord, NgR levels were highest at 55 days of gestation (not shown) where the entire white matter seemed to be positively labeled, as were many ventrally located neurons within the gray matter. By 61 days of gestation, spinal cord labeling was restricted largely to ventral neuronal cell bodies and positive staining was observed also in the dorsal root ganglia (Fig. 7G). This pattern of NgR protein expression parallels the distribution of NgR RNA at 63 days of gestation (9 weeks) recently reported (Josephson et al., 2002). Little staining was observed in the white matter, with the exception of the dorsal root entry zone, where projections from the dorsal root ganglia were clearly positive (Fig. 7G); the skeletal muscle was strongly positive (not shown). At the latest time point examined, 84 days of gestation, NgR staining was largely absent from the spinal cord, except for the pial membrane (Fig. 7H), but some positive staining was observed in dorsal root ganglia (not shown). A distinctive pattern of NgR expression, similar to that observed in the E5 chick eye, was noted in the 45-day retina (Fig. 7J), with the ganglion cells of the retina being positively labeled in peripheral but not central areas. The optic nerve was also NgR-positive.
Nogo is one of the most recently identified members of the reticulon gene family that is rapidly becoming established as being an ancient and widely distributed family of genes of yet unclear function. Gene sequences of Nogo isoforms have been obtained from human, rat, and cow and are highly conserved (Spillmann et al., 1998; Chen et al., 2000; Prinjha et al., 2000). The identification of numerous fish homologs of NgR again supports the notion of an evolutionary ancient signaling pathway (Klinger et al., 2003). We have demonstrated here the existence of Nogo-A in the chick, as clearly indicated by the detection of the long N-terminus sequence typical of this isoform, that is not present in any of the other members of the reticulon family so far identified. The conservation of the C-terminal domain within the reticulon family suggests a particularly crucial role for the transmembrane segment of the protein, possibly connected to the receptor-binding domain it contains. Although the inhibitory effects of the Nogo protein and its receptor on neurite outgrowth in vitro have been well established (Spillmann et al., 1997; Chen et al., 2000; GrandPre et al., 2000), the underlying physiological role of the protein in vivo remains to be determined. We show here that Nogo-A is developmentally regulated both in chick and human embryos, and it is first detected at stages of development when the chick spinal cord is able to regenerate. The conservation of this protein throughout evolution and its pattern of expression, together with that of its receptor, NgR, points to a crucial role of Nogo signaling during vertebrate development.
Nogo Is Developmentally Regulated
Our study shows that there is a switch during development in the cell types expressing Nogo-A. Whereas in adult rodent CNS Nogo-A is localized primarily in oligodendrocytes (Chen et al., 2000), during early development Nogo is associated with neurons. The onset of oligodendrocyte development in the chick and human has been previously determined by O4 immunostaining and occurs around E5–E6 in the chick and around 45 days of gestation in humans, whereas the onset of myelination is observed around E12 in chick and 20 weeks of gestation in humans (Ono et al., 1995; Tanaka et al., 1995; Hajihosseini et al., 1996; Grever et al., 1997). In this study, we have shown that, in the chick spinal cord, both Nogo-A transcript and protein can be detected at E3, a time when motor neurons are being generated and axon outgrowth begins, well before the onset of O4 expression. Indeed during early spinal cord development, the timing and localization of Nogo-A immunostaining coincides much more closely with axonal than oligodendrocyte development.
In human embryos, the onset of Nogo expression in the spinal cord could not be precisely determined because of the limited amount of embryonic stages available, but it clearly must occur between 28 days, when then spinal cord is Nogo-A–negative, and 45 days of gestation, when positive staining is observed. In human embryos, as in chick, Nogo is detected on axons, as indicated by our in vitro studies. However, unlike in the chick, where its expression is restricted to the white matter, some Nogo-positive staining is observed in the gray matter of human embryonic spinal cord.
We are confident that AS472 specifically cross-reacts with human Nogo-A, as in human protein samples reacted with a predominant single band of 220 kDa. In the chick, an additional band could be detected by the antibody. However, detection of Nogo-A transcripts in the spinal cord at E3, where AS472 labels motor neurons, together with the similarities between human and chick expression patterns, supports the notion that the Nogo-A protein is being specifically targeted by immunohistochemistry also in the chick. It should be noted that presence of additional bands, with lower and higher molecular weights than Nogo-A, is observed in other species, including the rat, where AS472 appears to be specific in immunohistochemical studies, as shown in experiments using the blocking peptide (Chen et al., 2000). Nevertheless, at present, we cannot exclude the possibility that a novel splice form of Nogo containing at least part of the Nogo-A–specific region exists in the chick. An alternative possibility might be that the high molecular weight band represents a strong complex of Nogo with other proteins such as NIMP, a mitochondrial protein recently shown to bind Nogo-A (Hu et al., 2002).
At early stages of human spinal cord development (41–61 dg), we have found no evidence of Nogo staining in O4-positive cells in vitro. At later stages, however (98 dg), a few double-labeled cells were present. Although these results may be biased by the low number of oligodendrocytes present in the cultures, they are consistent with the expression pattern observed in vivo, suggesting that, in human spinal cord, like in the chick, expression of Nogo in oligodendrocytes is a late event.
Expression of the Nogo protein in spinal cord neurons is consistent with observations made by other groups in the rat embryo (Huber and Schwab, 2000; Huber et al., 2002), and extend and complement a recent study on the human fetus reporting expression of Nogo mRNA in putative neurons at 9 weeks of gestation and in a subset of neurons in adult spinal cord (Josephson et al., 2001).
Nogo-A is developmentally regulated not only in the spinal cord but also in the peripheral nervous system, where it is found in the myenteric plexus and in peripheral nerves. In contrast, adult peripheral nerves do not express Nogo (Josephson et al., 2001).
At early stages of development, Nogo-A is expressed also in the notochord, the myotome, from which the skeletal muscle originates, and the epidermis of both chick and human embryos. Its expression is developmentally regulated both in skeletal and cardiac muscle, although some species differences are observed. We have shown that, in the chick, Nogo is detected in the myotome but not in the developing muscle, whereas in human embryos, Nogo-A is present in muscle fibers and in the heart at all the developmental stages examined, although it is not expressed in adult muscle in rat, mouse, or human (Chen et al., 2000; Goldberg and Barres, 2000; Josephson et al., 2001).
What Is Nogo's Role in Development?
Clues on the possible roles of Nogo during development, particularly in the chick, come from the distribution of its receptor, NgR, that binds a domain within the 66-amino acid region between the two putative transmembrane domains (Nogo-66). In the chick, NgR receptor and Nogo-A expression patterns appear to be complementary and mutually exclusive, the exception being their early expression in notochord and myotome. That at early stages of development NgR appears to be expressed on the surface of striated muscle, a target of peripheral nerve axons, and on neuronal cell bodies, together with the coordinated down-regulation of both ligand and receptor in neurons and muscles with development, are consistent with a potential signaling role of Nogo-A and NgR in target innervation and establishment of neural circuitry. This hypothesis is supported by recent localization of Nogo-A to synapses using electron microscopy (Liu et al., 2003). The generation of Nogo-deficient mice, however, has indicated that Nogo proteins are not essential for development of viable and fertile offspring (Zheng et al., 2003). Indeed the nature and extent of Nogo's influence on regeneration remains clouded, despite the creation of mutant mice lacking single/multiple Nogo isoforms. No improvement in the extent of axonal regeneration after spinal injury was reported by Zheng et al. (2003). However, similar strategies by other groups did result in modest enhancements of axonal repair (Kim et al., 2003; Simonen et al., 2003). If anything, the growing complexity of the Nogo signaling system has been eloquently demonstrated.
Understanding of the Nogo pathway must begin with elucidation of the principal mechanisms through which it influences specific cell types. It has been shown that Nogo-66 treatment of NgR-expressing neurites inhibits their outgrowth in culture. Therefore, on the basis of the in vitro data published so far (Chen et al., 2000; Goldberg and Barres, 2000; Fournier et al., 2001), it is tempting to assume that repulsion of axonal fibers, possibly in the context of deterring neurite contact with previously innervated tissues, is the key function of these proteins. In the system discussed here, however, it is Nogo not NgR that is present on the axons, whereas the receptor in vivo is mainly localized to neuronal cell bodies and muscle. Furthermore, in human spinal cord, as in the chick, expression of NgR protein (this study) and transcript declines with development and is virtually absent in the human adult cord (Josephson et al., 2002). Therefore, other possible roles for this signaling system, particularly during development, should be given careful consideration. For example, it has been proposed that another member of the reticulon family, NSP-C (now renamed Rtn1-C), may have a role in process extension (Senden et al., 1996), suggesting that Nogo-A expression on axons might play a role in axon elongation and possibly pathfinding. In addition, interactions between axonal Nogo and NgR might have a role in restricting spreading and movements of the NgR-expressing neurons upon reaching their target tissues, “locking” cells into specific positions. Nogo signaling, therefore, might be bidirectional and akin to other better characterized and complex signaling systems. Indeed, similarities between the ephrin/Eph receptor and Nogo/NgR families are already apparent: both can mediate axonal repulsion, and both share complementary expression domains within tissues (Mellitzer et al., 1999; O'Leary and Wilkinson, 1999; Feng et al., 2000). It may be possible to extrapolate clues regarding the function of Nogo by comparisons with ephrins and other inhibitory proteins.
Finally, it should be noted that the splice isoform Nogo-B recently has been shown to have proapoptotic activity, particularly in cancer cells, that is modulated by means of interactions with bcl-2 and bcl-xL (Tagami et al., 2000; Li et al., 2001). Mutant analysis has indicated that the apoptotic region resides in the second hydrophobic region and includes part of Nogo-66 (see Fig. 1). Because this region is common to all Nogo splice forms, it is possible that also Nogo-A has a similar proapoptotic activity and that it might play some role in programmed cell death during development, possibly by changing its cellular localization (translocation from ER to plasma membrane and vice versa) when finding/not finding its target.
N-Terminal Domain of Nogo-A Is Primarily Intracellular
The precise topology of the Nogo-A molecule in the plasma membrane is of crucial importance if deductions are to be made concerning the potential functional roles of Nogo during development. Previous studies have proposed differing orientations of the molecule in the plasma membrane. Specifically, the intracellular or extracellular location of the N-terminal domain that contains a putative functional site, has been a matter of debate (GrandPre et al., 2000; Huber and Schwab, 2000; Ng and Tang, 2002). The N-terminus contains an active region of Nogo that can collapse axons and inhibit fibroblast spreading, although it seems to evoke such response only when substrate bound (Fournier et al., 2001). In the course of this study, we have used two antibodies that specifically react with different peptides within the N-terminus of human Nogo. Although we cannot rule out entirely that there is some surface expression of Nogo-A's N-terminal domain in most cells at levels beyond the sensitivity of antibody detection, the significant reactivity observed in permeabilized human spinal cord cells and in a small percentage of live cells with both antibodies supports the view that, by and large, in these cultures the N-terminus region is not exposed on the cell surface. These antibodies, however, did not allow us to establish whether the Nogo-66 domain is mainly located on the ER or plasma membrane of the Nogo-positive axons. Other studies, however, have detected Nogo-A on the surface of cultured oligodendrocyte populations (Oertle et al., 2003). It should be noted, however, that the cultures used in this study reflect embryonic cell populations and the expression pattern of myelin-associated genes such as Nogo is likely to differ in adult stages.
Nogo-A and NgR Are Permissive for Spinal Cord Regeneration at Early Stages of Development
Nogo does not appear to play a negative role in regeneration at early stages of CNS development, as it is expressed before E12 in the chick, and it is not down-regulated in response to injury. The lack of inhibitory activity cannot be explained by the absence of NgR, as this receptor is also expressed at stages permissive for regeneration. Furthermore, as with Nogo-A, its levels of expression do not appear to be affected in response to injury (not shown). It is, therefore, unlikely that NgR expression during embryonic development may contribute to the transition from a regeneration-competent to -incompetent state.
It cannot be ruled out that another yet unidentified receptor mediating inhibition by means of a domain located in the N-terminus region is expressed only at later developmental stages nonpermissive for regeneration. However, given our evidence that in human embryonic neurons Nogo N-terminus domain is located intracellularly, this possibility seems less likely.
Because the inhibitory effects of Nogo become apparent only at later stages of development when its expression is down-regulated in axons and up-regulated in oligodendrocytes (Rubin et al., 1994; Huber and Schwab, 2000), it is conceivable that this switch in expression may be fundamental to the inhibitory effects of Nogo.
It is also conceivable that spillage of intracellular Nogo domains after spinal injury could contribute to failure of regeneration, possibly through a non–NgR-mediated pathway that is only active at later stages of embryonic development. Indeed, it has been shown recently that Nogo-A neurite growth inhibition is mediated by RhoA and RacC and does not necessarily require the presence of NgR (Niederost et al., 2002).
It should also be noted that, unlike the chick embryo, where we did not observe any down-regulation of Nogo, rat Nogo mRNA is significantly down-regulated in the adult spinal cord after a weight-drop injury (Josephson et al., 2001). Whether this finding is due to species difference, injury-type differences, or to developmental changes in function or regulation of Nogo remains to be determined.
In conclusion, we have shown that the Nogo signaling system is tightly regulated during development and, hence, likely to play key roles in the early development of the nervous system and striated muscle and that loss of regenerative potential is independent of Nogo-A and NgR expression per se but may rather be linked to changes in its cellular localization. The challenge lying ahead is to fully unravel the molecular components of this important new signaling pathway.
Tissues, Surgery, and Cell Culture
Fertilized White Leghorn eggs (Needle Farm, UK) were incubated at 37°C in a humidified forced flow incubator until the required developmental stage was reached. Spinal injuries were performed under aseptic conditions using an adaptation of the protocol described by Hasan and colleagues (1993). Briefly, eggs were windowed at E7, sealed with adhesive tape, and left to develop for another 4 days. E11 embryos were then lifted through the shell opening by using a glass hook, and the spinal cord was transected by using fine forceps. Eggs were then resealed and returned to the incubator for 24 hr.
Sections from human embryos between 28 days and 14 weeks of gestation, obtained under ethical approval, were provided by the Wellcome/MRC-funded Human Developmental Biology Resource. Embryonic human spinal cord tissue was obtained with consent from recent terminations. The tissue was dissected rapidly in cold L15 medium, and the spinal cord was finely chopped into small explants (approximate diameter 1 mm). These were seeded on laminin-coated glass coverslips in 24-well tissue culture dishes containing DMEM-glutamax (Gibco, UK) with 10% fetal calf serum (Sigma, UK), 1% penicillin/streptomycin (Sigma).
Tissues were fixed overnight in 4% paraformaldehyde (PFA) solution in phosphate-buffered saline (PBS) at 4°C. Fixed samples were embedded by immersion through a series of graded alcohols, Histoclear, and paraffin wax. Microtome sections were cut at 7 μm and floated out on poly-L-lysine–coated glass slides. Slides were dried overnight at 37°C before staining or storage. Sections were de-waxed by immersion in Histoclear and re-hydrated through a series of graded alcohols. To eliminate background labeling, endogenous peroxidase activity was blocked by immersion in 1% hydrogen peroxide solution for 15 min, and nonspecific protein interactions were blocked by incubation with 10% goat serum in PBS for 1 hr. Antibodies were diluted in 10% goat serum and applied overnight at 4°C. After washing in PBS containing 0.5% Triton X-100, appropriate secondary antibodies were applied. Secondary antibodies were diluted in 10% goat serum and incubation time was 1 hr at room temperature. Biotinylated secondary antibodies were detected by incubation with avidin–biotin horseradish peroxidase complex (ABC; Dako, UK) for 40 min followed by visualization with the substrate diaminobenzidine (Sigma, UK). Finally, the sections were counterstained with methyl green, dehydrated in butanol, and mounted with DPX (Fisher, UK).
For immunocytochemistry, cells grown on laminin-coated glass coverslips were labeled by application of 100 μl of primary antibody solution optimally diluted in L15 medium (Gibco, UK) for 30 min at room temperature. Coverslips were raised on small pedestals to facilitate ease of manipulation. Cells were fixed, either before or after primary antibody incubation by addition of 100 μl of acid alcohol (95% ethanol, 5% acetic acid) or PFA solution. Cells were then washed by immersion in three changes of L15 buffer, before application of fluorescently labeled secondary antisera, which again were diluted in L15 and incubated at room temperature for 30 min. The coverslips were finally washed in three changes of L15, rinsed in distilled water, and mounted on clean glass slides by using aqueous mountant (Citifluor, UK).
The primary antibodies used in this study were rabbit AS472 antiserum, previously shown to be Nogo-A–specific in the rat (Chen et al., 2000), was used 1:2000; rabbit anti-NgR, previously shown to react specifically with NgR in chick and mouse (Fournier et al., 2001; Wang et al., 2002), was used 1:3,000; goat anti–Nogo-A (S-19, used 1:100; Santa Cruz); goat anti-Nogo (N-18, used 1:100; Santa Cruz); mouse anti-neurofilament (RT-97, used 1:100; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA); mouse anti-O4 (MAB345, used 1:40, Chemicon); rabbit anti-GFAP (used 1:100, DAKO, UK).
Digital images were captured electronically by using a Zeiss Kontron ProgRes 3012 digital camera (Imaging Associates, Ltd., Thame, Oxon, UK), version 2.0 of the associated software and stored and labeled in Adobe Photoshop version 6.0 (Adobe Systems Europe, Edinburgh, UK). Fluorescent images were viewed under Zeiss axioplan microscope using fluorescent illumination and were digitally scanned by using a Hamamatsu digital camera (C4742-95, Hamamatsu Photonics KK, Japan) directly into Openlab software (version 3.03, Improvision, Ltd., www.improvision.com).
Carefully dissected spinal cords from chick embryos between E3 and E17 were homogenized in 1 ml of TRI-reagent (Sigma, UK) by using a syringe and narrow gauge needle. RNA was phenol/chloroform extracted following the manufacturer's instructions. cDNA was prepared by using M-MLV reverse transcriptase (Promega, UK) from 2 μg of RNA using either oligo-dT primers or random hexamers according to manufacturer's guidelines. PCR was performed using the following primer pairs: chick Nogo C-terminus: sense CGTTCTTGTTCACTAGTCCCAA, antisense TTGGAGTCTGATGTAGCTGTGT; chick Nogo-A N-terminus: (1) sense AGCTGAAGAAATGCAGCTCGAG, antisense, GTGTCACAGGTTTCAGAGTCTCTT, annealing temperature 59°C; (2) sense ATGCCTGAAGGTCTCACTCCTGAT, antisense GACTCCAAATAGGGCTTGCCAGTA; chick NgR: sense GACCTCAAAAGGCTGCAGAG, antisense TGAGAGGGTTGTTGGAGGAC; chick glyceraldehyde-3-phosphate dehydrogenase: sense CAGTGAGAAAGTCGGAGTCA, antisense GACACCCATCACAAACATGG (annealing temperature as above); human Nogo-A: sense GCTCTTCCTGCTGCATCTGAG, antisense TGCTCTCGATTTTACCTCCAGC; human NgR: sense CCAAGTGCTGCCAGCCAGAT, antisense TCAGCAGGGCCCAAGCACTGT; myelin basic protein (MBP), sense TTAGCTGAATTCGCGTGTGG, antisense GAGGAAGTGAATGAGCCGGTTA; human beta-actin: sense GTGGGCCGCTCTAGGCACCAA, antisense CTCTTTGATGTCACGCACGATTTC. PCR products were amplified by using Taq polymerase (Promega, UK) using the supplied protocol with annealing temperature of 55°C unless otherwise specified. PCR products were visualized on a 1.5% agarose gel containing ethidium bromide under ultraviolet light and imaged by using a Gel imaging system. PCR products of presumed chick Nogo sequence were excised and gel purified, before sequencing using the BigDye Terminator Cycle sequencing Ready reaction kit (Applied Biosystems, Warrington, UK).
Chick spinal cords were homogenized in lysis buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 0.1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate) containing protease inhibitors (1 μM sodium orthovanate, 100 μg/ml phenylmethylsulfonyl fluoride, 30 μl/ml aprotinin). The homogenates were spun at 12,000 rpm for 20 min, at 4°C. Supernatants were collected and stored at −70°C in small aliquots. Protein concentration in the supernatants was measured by using the BCA protein assay kit (Pierce, Chester, UK).
Proteins were separated by sodium dodecyl sulfate gel electrophoresis by using a 5% stacking gel and a 10% separating gel. A total of 50 μg of protein was loaded per lane. The proteins were then transferred to a Hybond-C membrane (Amersham, Buckinghamshire, UK) by semi-dry electroblotting. The filters were placed overnight in blocking solution (5% Marvel, 0.1% Tween-20 in PBS) and then incubated with AS472 (1:2,000) in blocking solution at 4°C for 1 hr. Membranes were washed three times in blocking solution, and then incubated for 30 min with a peroxidase-conjugated goat anti-rabbit antibody (DAKO, UK) diluted 1:1,500 in blocking solution. After three washings in PBS, the positive signal was detected with the enhanced chemiluminescence reagent kit (ECL, Amersham, Buckinghamshire, UK).
We thank Martin Schwab and Andrea Huber for the AS472 antibody, for sharing unpublished data, and for useful discussion; and Stephen Strittmatter for providing the anti-Nogo receptor antibody. We also thank Caroline Paternotte for assistance with DNA sequencing.