Direct observation of myelination in vivo in the ... - Semantic Scholar

Brain (1997), 120, 2071–2082

Direct observation of myelination in vivo in the mature human central nervous system A model for the behaviour of oligodendrocyte progenitors and their progeny Samuel F. Hunter,1,2,* Jacqueline A. Leavitt3 and Moses Rodriguez1,2 Departments of 1Neurology, 2Immunology and 3Ophthalmology, Mayo Clinic, Rochester, USA

Correspondence to: Dr Moses Rodriguez, Gug. 428, 200 1st St S.W., Rochester, MN 55905, USA *Present address: MS Research Laboratory, Vanderbilt Stallworth Rehabilitation Hospital, 2201 Capers Avenue, Room 1222, Nashville, TN 37212, USA

Summary We studied patches of CNS myelin in human retina in vivo to determine the pattern of myelination and the local influence of axons. We analysed the position, area and thickness of the nerve-fibre layer in 60 patches of retinal myelin in 47 eyes (in 37 adults and two adolescents). Five patches in four eyes were studied serially over 6–11 years. Nerve-fibre layer thickness was obtained from an atlas of primate retina, and volumes of myelinated tissue were then estimated for each patch. Retinal myelination occurred in three patterns: thick patches contiguous with the optic disc (type I); thin, striated patches detached from the disc (type II); or massive myelination of the posterior pole associated with severe amblyopia (type III). The papillomacular bundle did not myelinate in types I and II and was relatively spared in type III patches, suggesting that migratory oligodendrocyte progenitors are not supported by these axons. The local

nerve-fibre layer determined patch size, and quantal myelination was evident with modal peaks of patch volume at 0.16 and 0.64 mm3. Myelination advanced at patch edges when observed over time, consistent with the hypothesis that new oligodendrocytes are produced in adulthood. We propose a theoretical model where patches of retinal myelination are the clonal progeny of a few oligodendrocyte progenitors exhibiting two different behaviours. First, a highly migratory, nonmyelinating progenitor uses larger, phylogenetically older axons as the substrate for movement. Secondly, a more mature progenitor generates myelinating oligodendrocytes well into adult life, but traverses only short distances. Using this data, we can estimate the number of oligodendrocytes in these clones and population doubling-time. This study supports a role for axon-derived signals in the regulation of human oligodendrocyte progenitor behaviour and myelination in vivo.

Keywords: oligodendrocyte progenitor; development; retina; regeneration; multiple sclerosis

Introduction Oligodendrocyte regeneration following injury in human central nervous system remains a pivotal issue in the treatment of demyelinating diseases, especially multiple sclerosis (Hunter and Rodriguez, 1995; Hunter et al., 1997). Studies of adult human oligodendrocyte progenitors in vivo have not previously been performed. New strategies for myelin repair require additional information about the potential for the proliferation and migration of these cells. We have found that patches of myelinated retina in humans provide a unique model for the quantitative study of human oligodendrocyte © Oxford University Press 1997

development and the interrelationship of this lineage with axons. We propose the following hypothesis. Patches of retinal myelin are formed by the clonal progeny of a few founding oligodendrocyte progenitors which enter the retinal nervefibre layer, migrate and expand under local axonal influence. The normally unmyelinated primate retina provides a paradigm for the behaviour of oligodendrocyte lineage cells and an opportunity to observe myelination. The retinal nervefibre layer consists of ganglion cell axons traversing the

2072

S. F. Hunter et al.

superficial retina to converge at the optic nerve head, where they turn posteriorly, passing through the dense astrocytic processes of the lamina cribosa. This structure probably prevents retinal myelination (Berliner, 1931), presumably by preventing entry of migratory oligodendrocyte progenitors (ffrench-Constant et al., 1988). Consistent with this theory, the normally unmyelinated retina of rats also myelinates when immature oligodendrocyte lineage cells are engrafted (Huang et al., 1991) and posterior to the lamina cribosa all ganglion cell axons become myelinated by oligodendrocytes. Proliferative and migratory oligodendrocyte progenitors produce oligodendrocytes in the rat optic pathways (Raff et al., 1983; ffrench-Constant and Raff, 1986). Variations of axonal anatomy, organization and thickness in the retinal nerve-fibre layer present an opportunity to study neuron–oligodendrocyte interactions, as neurons and axons play a crucial role in vivo in development of oligodendrocytes from undifferentiated, proliferative progenitors (Fulcrand and Privat, 1977; David et al., 1984; Barres and Raff, 1993). A specialization of the primate retinal nerve-fibre layer is the papillomacular bundle, which traverses from the macular region to the temporal disc margin (Fig. 1A; Hoyt and Tudor, 1963). This region is bounded inferosuperiorly by the arcuate fasciculi and temporally by the temporal raphe (Vrabec, 1966). The nerve-fibre layer is thickest (200 µm) where the arcuate fasciculi converge on the disc (Fig. 1A and B), and larger laminar pores admit these bundles to the optic nerve (Quigley and Addicks, 1981; Jonas et al., 1991). The nervefibre layer declines exponentially in thickness with distance from the disc (Fig. 1B; Radius, 1980; Ogden, 1983) and is partitioned by Mu¨ller glia into ophthalmoscopically visible fascicles which radiate from the optic disc (Hoyt et al., 1972; Radius and Anderson, 1979). We present quantitative data which support our hypothesis and suggest a theoretical model for the behaviour of human oligodendrocyte progenitors and their regulation by axonderived signals. Our findings also suggest a three-phase model of human myelination. We provide estimates for the size and growth of a clonal population of oligodendrocytes derived from founding oligodendrocyte progenitors in the CNS.

Methods Identification of cases Using the photographic index in the Department of Ophthalmology at the Mayo Clinic, we identified photographs of retinal myelin taken from 1972 to 1996. A total of 60 discrete patches were studied in 47 eyes of 37 adults and two adolescents. The characteristics of the study population are given in Table 1. Medical histories were reviewed to determine the reason for referral, the best corrected visual acuity and ocular comorbidity, and to confirm the patch position. In cases of serial photographs, the patients were either followed for research studies (diabetics with mild

background retinopathy) or choroidal nevi. All photographs were reviewed by a neuro-ophthalmologist (J.A.L.) to confirm the diagnosis of myelinated fibres. Patches were typed and classified as described in Table 1.

Photography Retinal photography was performed with either a 30° field (32.5; Zeiss Fundus Camera) or a 60° field (31.7; Canon CF507A Fundus Camera) onto 35-mm colour positive film.

Digital image analysis The 35-mm colour positives were digitized in colour at high resolution (6003800 pixels for 24332 mm image) using a LeafScan 45 scanner (Leaf Instruments, Southborough, Mass., USA) at 25 pixels per mm of film. Usually a pixel corresponded to a retinal distance of 16 µm (with 30° view) or 23.5 µm (with 60° view) in both vertical and horizontal dimensions. Higher resolution of smaller areas of the photographs was used for studying serial images of the same patch (6.1–8.8 µm retina per pixel). Interactive planimetric measurements were made from the green channel of the digitized image, permitting excellent discrimination of retinal myelin and the normal retinal nerve-fibre layer. Measurements were performed by a single investigator (S.F.H.) using KS400 image analysis software (Kontron Electronik Gmbh, Munich) on a Pentium PC with Windows 3.1. The area of each separate patch was determined by tracing the outline of the patch, and pixels were converted to square millimetres. When multiple retinal patches were present, they were analysed separately, but contiguous areas of myelination were not divided for analysis. A distance of the patch centroid from the disc edge was computed when possible, and estimated when not directly measured. For measurements of serial photographs, three or six independent measurements were made of each photograph with resulting coefficients of variation of 2–10%. For these cases, measurements were made for both lightly and densely myelinated regions. The former were marked at the outermost perceptible change of the nerve-fibre layer and the latter as the area which partially or completely obscured underlying retinal structures. Least-squares regression lines were calculated using an iterative Marquardt–Levenburg method through the serial area measurements for both the lightly and heavily myelinated regions. The fit was constrained so that both regression lines had a common origin on the time axis. This provided an unbiased extrapolation of calculated age at initiation of myelination. We used a second, intensity-based method to confirm patch growth. Selected digitized images of patches at early and late timepoints were registered and pre-processed to produce identical intensity of highlights, shadows and mid-range tones of the green-filtered images. A sharp border to the patch edge was defined by adjusting the black level to zero and the contrast to a level which

Human oligodendrocyte progenitors in vivo

2073

Fig. 1 Anatomy and nerve-fibre layer thickness of the retina. (A) The relevant retinal anatomy for this study is labelled in an idealized left eye. (B) The template used for estimating nerve-fibre layer thickness, modified and redrawn from Ogden (1983) with kind permission from Elsevier Science Ltd, Oxford, UK. The nerve-fibre layer thickness in various positions is given in micrometres. Note the 40-fold decline in nerve-fibre layer thickness from the disc to the edges of the vascular arcades. The bar represents 1 mm.

produced a binary image with an edge between the observed lightly and densely myelinated zones. This edge was overlaid on both the early and late timepoint images to demonstrate advancement of myelination.

Estimates of patch volume Patch volume was estimated as the product of the measured retinal area and the estimated nerve-fibre layer thickness in the patch position. We used Ogden’s published atlas for

2074

S. F. Hunter et al. Table 1 Patterns of myelination and characteristics of study population. Patch types and subtypes

Patches (n)

Eyes (n)

Patients

Age in years (median)

Type I, all patches ,50% disc margin ù50% disc margin

32 17 15

26 15 11

11F, 10M 7F, 7M 5F, 4M

24–81 (62) 24–81 (62) 25–73 (62)

Type II, all patches Arcuate fasciculus Raphe region Other regions (nasal)

24 4 5 15

17 4 5 8

4

4

2F, 2M

11–59 (16)

60

47

17F, 22M

11–81 (55)

Type III, all patches Total

4F, 1F, 2F, 2F,

10M 3M 2M 5M

18–73 30–70 49–70 18–73

(56) (57) (59) (53)

Type I 5 myelin contiguous with optic nerve head, no symptomatic amblyopia; Type II 5 myelin discontinuous with optic nerve head; Type III 5 extensive myelination (at optic nerve head and distant) with symptomatic amblyopia. Examples of each patch type are illustrated in Figs 2 and 3. The number of eyes and patients in the subclassifications do not add to the total for the patch type because some individuals have bilateral patches and eyes in different subclasses (see text). Type I patches were subclassified according to the fraction of the disc margin which was obscured by myelination. Type II patches were subclassified according to their location. F 5 female; M 5 male.

primate retina (Fig. 1B; Ogden, 1983) to estimate the nervefibre layer thickness. This extensive map has evenly spaced determinations, making it ideal for mapping nerve-fibre layer thickness of patches. The patch was sketched over the map, and values at covered points were averaged to yield a mean estimated thickness. Thus, the estimated thickness and volume of patches were equivalent to those of an unmyelinated region of nerve-fibre layer in the normal primate eye. The number of oligodendrocytes per patch was estimated as follows. Nerve-fibre layer volume (µm3) was multiplied by the fraction of axonal volume (0.6). This result can be considered as a product of axonal cross-sections (in µm2) and axonal length (in µm) and converted to total axon length (in µm) by dividing by an average axon cross-section of 1 µm2. The total axon length was divided by 175 µm of axon length per internode to yield internodes per patch, and this figure is divided by 20 internodes per oligodendrocyte to yield the number of oligodendrocytes per patch.

Results Three distinct myelination patterns occur Retinal myelination occurred in only three patterns (Table 1; Figs 2 and 3): contiguous with the optic disc (type I), detached from the optic disc (type II) and as massive retinal myelination associated with severe amblyopia (type III). Cases with type I and II patches had similar median ages at time of photography, were asymptomatic and were noted either on a routine examination or due to other ocular pathology. Those with type III patches had a younger median age, consistent with symptomatic amblyopia, and demonstrated quantitatively greater myelination in a pattern resembling both type I and II. Type I and type II patches were further subclassified to simplify analysis (Table 1). Bilateral patches occurred in 7 out of 39 patients (18%) and

were always of the same type (I or II) in both eyes. No bilateral type III patches occurred. Multiple patches were present in six eyes; three eyes had two type I patches, one eye had three type I patches, and two eyes from the same individual had four and five type II patches.

The nerve-fibre layer determines the appearance and area of patches Composite plots of retinal patches demonstrated a consistent pattern (Fig. 2). Although the papillomacular bundle contains 25% of all retinal axons (Potts et al., 1972), myelin completely spared this region in the 56 type I and II patches, and relatively spared this region in the other four patches (type III). All other regions had the potential to become myelinated, although type II patches spared the more proximal, and thicker, arcuate fasciculi. Instead, type II patches occurred in nerve-fibre layer with an estimated thickness of ,30 µm (Fig. 4A). Patches of the same type in similar locations were similarly sized (Fig. 4B). For type I and II patches which did not circumscribe the optic nerve head, the area correlated with increased distance of the patch centroid from the disc edge (r 5 0.69, Spearman’s rank correlation; Fig. 4C). Nerve-fibre layer thickness correlated inversely (r 5 20.65, Spearman’s) with patch area for both type I and type II patches (Fig. 4D). Thus, nerve-fibre layer thickness determined in part the area of the patch, and specific quanta of myelinated tissue could then be estimated.

Quantal volumes of retinal myelination We estimated the equivalent volume of normal retina which was myelinated for all 60 patches using local nerve-fibre layer thickness of an unmyelinated primate retina. The probability distribution histogram for patch volume revealed


Fig. 2 Positions of 60 patches of retinal myelin. (A and B) Type I and II patches (see Table 1 and text) have been plotted onto a retinal template for 22 right (A) and 21 left eyes (B). Increased density of the shading represents increased overlap of separate patches. Note the sparing of the papillomacular bundle in type I and II patches. (C) The type III patches (or a mirror image) are plotted for four eyes. The papillomacular bundle is only relatively spared.

2075

2076

S. F. Hunter et al.

Fig. 3 Examples of retinal myelination patterns. (A) Two asymptomatic type I patches involving optic disc in 52-year-old with a contralateral choroidal nevus. (B) Large, asymptomatic type II patch in the distal arcuate fasciculus and temporal raphe of a 58-year-old. (C) Asymptomatic type II patches in nasal retina of a 44-year-old with multiple bilateral patches. (D) Montage illustrating a type III patch in a massively myelinated, amblyopic eye of a 26-year-old with visual acuity recorded as ‘light perception only.’ The bars represent 1 mm.

discrete peaks and a multimodal pattern (Fig. 5A). Peaks were localized at volumes which are powers of two (or sums thereof) from the prominent mode at 0.64 mm3. The second largest peak occurred at 0.12–0.16 mm3. The smallest patch volumes were type II patches (Fig. 5B). Age also had an effect, as 13 out of 15 of the smallest patch volumes (,0.4 mm3) occurred in individuals who were ,45 years old.

New myelin is formed in adulthood Four eyes with type I and II patches had not been found to have myelin on previous dilated examinations 1–44 (median

13) years prior to photography, again suggesting possible patch origin and growth in adulthood. We studied serial photographs of five type I and II patches over 6–11 years in four adult eyes. Progressive enlargement was observed in all patches, even in the sixth decade of life (Figs 6 and 7). Linear growth of patch area was observed in both densely and lightly myelinated zones. Extrapolation of the growth patterns by least-squares regression to a common point on the time axis indicated an origin in childhood, adolescence, or adulthood, depending on the patch (Fig. 6). An intensitybased method was used to confirm advancement and intensification of myelination in selected cases (Fig. 7C and D).


2077

Fig. 5 Quantal myelination of human retina. The equivalent volume of unmyelinated nerve-fibre layer was determined for each patch as described in Methods. (A) The histogram for patch volumes demonstrates discrete peaks; the pattern is neither a normal nor uniform distribution. Peaks occur near powers of two from the principal mode, or at sums of powers of two. When a peak occurs at such a volume, it has been marked with the patch volume and arrow. (B) The volumes of individual patches are plotted by type and location. See Fig. 4 legend for symbol key.

Fig. 4 Retinal region and nerve-fibre layer determine patch characteristics. (A and B) Patch properties by type and subclassification. (A) The thickness of involved nerve-fibre layer for type II patches is much less than type I patches. (B) Patch areas within groups form clusters. (C) Patch areas increase as the patch centre moves farther from the disc edge. Note that many larger type I patches are centred about the optic nerve head and this distance is zero. Type III patches are omitted for clarity. (D) Estimated nerve-fibre layer thickness correlates inversely with patch area. Circles represent type I patches covering ,50% of disc margin (I ,50); squares represent type I patches covering ù50% of disc margin (I .50); triangles represent type II patches partially involving arcuate fasciculus (II-AF); inverted triangles represent type II patches involving temporal raphe (II-R); diamonds represent type II patches involving other regions (II-O); hexagons represent type III patches. Shaded symbols represent patches for which some edges were not observed in the photograph.

Discussion The asymptomatic developmental variant of limited retinal myelination in humans probably occurs because of a minor failure to prevent entry of oligodendrocyte lineage cells. This function is normally performed by a barrier at the distal end of the optic nerve, either the lamina cribosa (Berliner, 1931) or a similar anatomical structure (ffrench-Constant et al., 1988). These patches seem to represent a natural experiment which illustrates developmental mechanisms important in myelination.

Our quantitative study of CNS myelin patches in human retina supports the following behaviours of oligodendrocyte lineage cells. (i) Two different migratory behaviours for oligodendrocyte progenitors occur in development. (ii) Certain axons (papillomacular bundle) cannot support oligodendrocyte progenitor migration. (iii) New myelinating oligodendrocytes are generated slowly even in adulthood. (iv) Myelin is formed by the clonal progeny of a small founding unit of oligodendrocyte progenitors. These findings may have important implications for human CNS development and for myelin repair. Several anecdotal reports have documented the appearance or enlargement of myelin patches in children (Baarsma, 1980; Aaby and Kushner, 1985; Ali et al., 1994), whereas patches in adults have previously been thought to be stable (Straatsma et al., 1981). Retinal myelination usually occurs asymptomatically as a developmental variant in ,1% of human eyes, although ~10% of these cases show extensive myelination with ipsilateral myopia, amblyopia and strabismus (Straatsma et al., 1979, 1981; Kodama et al., 1990). Even though human retinal myelin is not contiguous with the optic nerve myelin, oligodendrocytes produce the myelin found in both the human and the normally myelinated rabbit retina (Berliner, 1931; Straatsma, 1981; Reichenbach, 1988; Yanoff and Fine, 1989). Retinal myelin degenerates following axonal injury to the optic pathways (Sachsalber, 1905; Bachmann, 1921; Schachat and Miller, 1981); two of our patients had partial regression due to optic nerve injury. Fascicular atrophy of myelin has been observed following optic neuritis (Sharpe and Sanders,

2078

S. F. Hunter et al.

Fig. 6 Expansion of retinal myelin in adulthood. Areas of five patches in four eyes are plotted against agein years at the time of the photograph. Patients 1 and 3 were diabetics requiring insulin with mild background retinopathy. Patient 2 was followed for a choroidal nevus in the left eye, and two patches were detected incidentally in the right eye. Two points (mean 6 SEM for three or six determinations) are plotted for each photograph. Outermost patch edges (open symbols) were delineated by minimally perceptible alteration in the nerve-fibre layer. The heavily myelinated region of patch (closed symbols) obscured retinal structures. Linear regression lines (see Methods) predict an age where linear growth patterns converge. Progressive enlargement occurred in all patches. Note the extrapolated origin in adulthood for some patches.

1975), similar to nerve-fibre layer loss in the unmyelinated retinas of multiple sclerosis patients (Frisen and Hoyt, 1974). Our study indicates that adulthood may not be an absolute obstacle to oligodendrocyte production in human optic pathways, although the spread of myelination is quite slow. Others have failed to notice the growth of patches in adults, possibly due to inadequate sampling and magnification. The progressive myelination of retina may be analogous to progressive myelination of the intracortical plexus and reticular formation throughout human adolescence and adulthood (Yakovlev and Lecours, 1967). Slow growth at the distal edge and myelination distant from the optic nerve head (type II patches) support the founding and growth of patches by a small number of oligodendrocyte progenitors exhibiting two behaviours. First, some founding progenitors migrate prior to generating oligodendrocytes, and may be the equivalent of the highly proliferative, motile, ‘neonatal’ rat oligodendrocyte (O-2A) progenitor (Raff et al., 1983, 1988; Noble and Murray, 1984; Small et al., 1987; Curtis et al., 1988; Hunter and Bottenstein, 1989). Secondly, the appearance of new myelin at patch edges requires persisting progenitors which slowly generate new oligodendrocytes, but migrate only short distances along or across unmyelinated axons. These stable, slowly differentiating ‘mature brain’ progenitors migrate and proliferate much less than the ‘neonatal’ phenotype, but emerge from populations of neonatal progenitors in vitro (ffrench-Constant and Raff, 1986; Hunter and Bottenstein, 1989, 1990, 1991; Wolswijk and Noble, 1989; Wren et al., 1992). Mature brain progenitor behaviour may be partially determined by trophic signals during development (Hunter and Bottenstein, 1991; Tontsch et al., 1994), and in vitro evidence suggests that these cells divide assymetrically to

renew the progenitor cell and to produce differentiated oligodendrocytes (Wren et al., 1992). Although others (Armstrong et al., 1992; Gogate et al., 1994) theorize that new oligodendrocytes in adult humans arise from a reserve of post-mitotic ‘pre-oligodendrocytes’, increases in retinal myelin over many years are most likely explained by birth of new oligodendrocytes. The potential for two progenitor behaviours may already exist in human foetal life, since some second-trimester, human oligodendrocyte progenitors differentiate quickly into oligodendrocytes in vitro, while others remain undifferentiated (Kennedy and Fok-Seang, 1986). However, proliferative oligodendrocyte lineage cells in adult humans are either absent (Armstrong et al., 1992; Gogate et al., 1994) or ,1% of this lineage (Prabhakar et al, 1995; Scolding et al., 1995; Hunter and Rodriguez, 1996). Whether this tiny fraction of cells produces new oligodendrocytes has been a crucial issue, since mature human brain oligodendrocyte lineage cells do not make myelin in the demyelinated rat spinal cord (Targett et al., 1996), whereas embryonic human CNS cells produce normal myelin in a dysmyelinating mutant (Gumpel et al., 1987). Retinal myelin patches also show that a subpopulation of axons directs either migration, proliferation or survival of oligodendrocyte progenitors. Our data imply that the papillomacular bundle does not support oligodendrocyte progenitors in the retina, as shown by absence of retinal myelin from this distinct axonal subpopulation (Potts et al., 1972). The papillomacular bundle has a very small proportion of large axons (median 0.5 µm diameter) in contrast to nasal bundles or arcuate fasciculi (0.8–0.9 µm; Ogden, 1984). Oligodendrocyte progenitors best align their processes on grooved substrates which are of comparable size to larger,


2079

Fig. 7 Examples of new myelin production in adults. (A and B) Patient 1: type II patch (inferior temporal) in 1983 (age 20 years, A) and 1992 (age 29 years, B). The local nerve-fibre layer thickness was estimated to be 17 µm and volume at the final measurement was 0.17 mm3. Spread of both lightly and densely myelinated regions is apparent. (C and D) Patient 2: type I patch, left eye 1985 (age 52 years, C) and 1992 (age 59 years, D), illustrating the intensity-based method for determining patch growth (see Methods). In C (heavy line) the myelination front has been delineated and it is reproduced in (D). The new patch edge (the isobar of similar intensity) is also marked (D; light line). This method marks a myelination front which is intermediate between the two extremes depicted in Fig. 6. The nerve-fibre layer thickness was estimated to be 145 µm and volume 0.41 mm3. The green filtered images are shown. The bars represent 100 µm.

unmyelinated axons (Webb et al., 1995). In addition, phylogenetically younger regions, such as the papillomacular bundle, may become myelinated later as they do elsewhere in the CNS (Minckler and Boyd, 1968). It is unlikely that the lamina cribosa prevents oligodendrocyte progenitors from entering the temporally placed macular fibres more efficiently, since temporal and nasal laminar pores are similarly sized (Quigley and Addicks, 1981; Jonas et al., 1991), and the nasal disc and retina frequently become myelinated whereas the temporal macular fibres do not. Local axonal mass could be responsible for the switch of progenitor migratory behaviour. Type II patches appear to

be founded by a premyelinating, migratory cell which stops its migration and begins myelination upon entry into thinner nerve-fibre layer. This behaviour may mimic the cessation of proliferation and differentiation of neonatal rat oligodendrocyte progenitors caused by loss of trophic signals in vitro (Raff et al., 1988; Hunter and Bottenstein, 1989). Nerve-fibre layer thickness (axonal mass) seems to have an ongoing influence on patch development, since the smallest patch volumes occurred in thin nerve-fibre layer of younger individuals. Similar patch sizes in similar retinal regions indicate that a developmental mechanism limits the amount of myelinated

2080

S. F. Hunter et al.

tissue. The nature of the defect in the barrier in the distal optic nerve is unclear but appears limited in most cases; molecular signals have been described which inhibit oligodendrocyte progenitor migration (Kiernan et al., 1996). Estimates of the volume of involved nerve-fibre layer show a striking pattern consistent with quantal myelination. These findings support our hypothesis for retinal myelin origin from only a few founding oligodendrocyte progenitors, implying that ultimate patch size is largely determined early in development. Our volume estimates are not of the exact volume of the myelinated patch, since we cannot account for expansion of the nerve-fibre layer secondary to myelination, partially myelinated fascicles or minor variations in Mu¨ller cell and astrocyte volume (Ogden, 1984). However, we can compare patch volumes of the equivalent nerve-fibre layer in a normal primate eye, assuming that patches contain similar fractions of myelinated axons and glia. Based on our observations, we propose that human CNS myelination occurs in three phases. First, dissemination of the oligodendrocyte lineage takes place as the premyelinating, migratory oligodendrocyte progenitors move longitudinally on larger axons. These first progenitors may cease migration when a local axonal signal declines below a critical threshold. In the second phase progenitors generate oligodendrocytes responsible for early myelination and mature brain progenitors. In the third phase, consolidation of myelination occurs as the mature brain progenitors migrate only short distances, slowly producing myelinating oligodendrocytes. This scheme is consistent with the small degree of myelination in foetal life (Yakovlev and Lecours, 1967; Weidenheim et al., 1996), rapid myelination in the first postnatal year and subsequent slow progression of myelination, reaching essentially an adult pattern at five years of age (van der Knaap and Valk, 1995), although histological myelination contines to progress in adulthood in some regions (Yakovlev and Lecours, 1967). These volumes of involved nerve-fibre layer permit estimates of clonal progeny of oligodendrocytes from a single progenitor in vivo. Using the smallest type II patch volumes (0.06 mm3), an astroglial volume content of 40% (Ogden, 1984), 20 internodes per oligodendrocyte, 175 µm per internode (Butt and Ransom, 1989) and an average axon cross-sectional area of 1 µm2 (Potts et al., 1972), we estimate that there are 104 oligodendrocytes in these patches (see Methods). The smallest type I patches generate a fourfold greater figure, and the modal patch volume (0.64 mm3) represents ~105 oligodendrocytes. Patch growth is probably the consequence of multiple proliferative progeny which arise from founding cells. Assuming a neglible loss of oligodendrocytes and growth of the type II patch (Fig. 6), we estimate a birth rate of 103 oligodendrocytes per year with a population doubling-time of ~10 years. Assymetric progenitor division every 7 days would mean ~0.2% of the oligodendrocyte population are progenitors.

Acknowledgements We wish to thank Thomas Link and Jay Rostvold for technical assistance with the fundus photographs. S.F.H. is supported by a Clinician Investigator award from Mayo Foundation. This work was supported in part by a grant from Research to Prevent Blindness, New York.

References Aaby AA, Kushner BJ. Acquired and progressive myelinated nerve fibers. Arch Ophthalmol 1985; 103: 542–4. Ali BH, Logani S, Kozlov KL, Arnold AC, Bateman B. Progression of retinal nerve fiber myelination in childhood. Am J Ophthalmol 1994; 118: 515–7. Armstrong RC, Dorn HH, Kufta CV, Friedman E, Dubois-Dalcq ME. Pre-oligodendrocytes from adult human CNS. J Neurosci 1992; 12: 1538–47. Baarsma GS. Acquired medullated nerve fibres. Br J Ophthalmol 1980; 64: 651. Bachmann R. Schwund markhaltiger Nervenfasern in der Netzhaut nach Embolie der Art. centralis retinae. Arch f. Ophth Berl 1921; 17: 10–4. Barres BA, Raff MC. Proliferation of oligodendrocyte precursor cells depends on electrical activity in axons. Nature 1993; 361: 258–60. Berliner ML. Cytologic studies on the retina. I. Normal coexistence of oligodendroglia and myelinated nerve fibers. Arch Ophthalmol 1931; 6: 740–51. Butt AM, Ransom BR. Visualization of oligodendrocytes and astrocytes in the intact rat optic nerve by intracellular injection of lucifer yellow and horseradish peroxidase. Glia 1989; 2: 470–5. Curtis R, Cohen J, Fok-Seang J, Hanley MR, Gregson NA, Reynolds R, Wilkin GP. Development of macroglial cells in rat cerebellum. I. Use of antibodies to follow early in vivo development and migration of oligodendrocytes. J Neurocytol 1988; 17: 43–54. David S, Miller RH, Patel R, Raff MC. Effects of neonatal transection on glial cell development in the rat optic nerve: evidence that the oligodendrocyte-type 2 astrocyte cell lineage depends on axons for its survival. J Neurocytol 1984; 13: 961–74. ffrench-Constant C, Raff MC. Proliferating bipotential glial progenitor cells in adult rat optic nerve. Nature 1986; 319: 499–502. ffrench-Constant C, Miller RH, Burne JF, Raff MC. Evidence that migratory oligodendrocyte-type-2 astrocyte (O-2A) progenitor cells are kept out of the rat retina by a barrier at the eye-end of the optic nerve. J Neurocytol 1988; 17: 13–25. Frisen L, Hoyt WF. Insidious atrophy of retinal nerve fibers in multiple sclerosis. Arch Ophthalmol 1974; 92: 91–7. Fulcrand J, Privat A. Neuroglial reactions secondary to Wallerian degeneration in the optic nerve of the postnatal rat: ultrastructural and quantitative study. J Comp Neurol 1977; 176: 189–222. Gogate N, Verma L, Zhou JM, Milward E, Rusten R, O’Connor M, et al. Plasticity in the adult human oligodendrocyte lineage. J Neurosci 1994; 14: 4571–87.

Human oligodendrocyte progenitors in vivo Gumpel M, Lachapelle F, Gansmuller A, Baulac M, Baron van Evercooren A, Baumann N. Transplantation of human embryonic oligodendrocytes into shiverer brain. Ann NY Acad Sci 1987; 495: 71–85. Hoyt WF, Tudor RC. The course of parapapillary temporal retinal axons through the anterior optic nerve: a Nauta degeneration study in the primate. Arch Ophthalmol 1963; 69: 503–7. Hoyt WF, Schlicke B, Eckelhoff RJ. Fundoscopic appearance of a nerve-fibre-bundle defect. Br J Ophthalmol 1972; 56: 577–83. Huang PP, Alliquant B, Carmel PW, Friedman ED. Myelination of the rat retina by transplantation of oligodendrocytes into 4-day-old hosts. Exp Neurol 1991; 113: 291–300. Hunter SF, Bottenstein JE. Bipotential glial progenitors are targets of neuronal cell line-derived growth factors. Brain Res Dev Brain Res 1989; 49: 33–49.

2081

Ogden TE. Nerve fiber layer of the primate retina: thickness and glial content. Vision Res 1983; 23: 581–7. Ogden TE. Nerve fiber layer of the primate retina: morphometric analysis. Invest Ophthalmol Vis Sci 1984; 25: 19–29. Potts AM, Hodges D, Shelman CB, Fritz KJ, Levy NS, Mangnall Y. Morphology of the primate optic nerve. 3. Fiber characteristics of the foveal outflow. Invest Ophthalmol 1972; 11: 1004–16. Prabhakar S, D’Souza S, Antel JP, McLaurin J, Schipper HM, Wang E. Phenotypic and cell cycle properties of human oligodendrocytes in vitro. Brain Res 1995; 672: 159–69. Quigley HA, Addicks EM. Regional differences in the structure of the lamina cribrosa and their relation to glaucomatous optic nerve damage. Arch Ophthalmol 1981; 99: 137–43. Radius RL. Thickness of the retinal nerve fiber layer in primate eyes. Arch Ophthalmol 1980; 98: 1625–9.

Hunter SF, Bottenstein JE. Growth factor responses of enriched bipotential glial progenitors. Brain Res Dev Brain Res 1990; 54: 235–48.

Radius RL, Anderson DR. The histology of retinal nerve fiber layer bundles and bundle defects. Arch Ophthalmol 1979; 97: 948–50.

Hunter SF, Bottenstein JE. O-2A glial progenitors from mature brain respond to CNS neuronal cell line-derived growth factors. J Neurosci Res 1991; 28: 574–82.

Raff MC, Miller RH, Noble M. A glial progenitor cell that develops in vitro into an astrocyte or an oligodendrocyte depending on culture medium. Nature 1983; 303: 390–6.

Hunter SF, Rodriguez M. Multiple sclerosis: a unique immunopathological syndrome of the central nervous system. [Review]. Springer Semin Immunopathol 1995; 17: 89–105.

Raff MC, Lillien LE, Richardson WD, Burne JF, Noble MD. Platelet-derived growth factor from astrocytes drives the clock that times oligodendrocyte development in culture. Nature 1988; 333: 562–5.

Hunter SF, Rodriguez M. In vitro myelin antigen expression and responses to trophic factors indicate at least two types of oligodendroglia in mature human brain[abstract]. Neurology 1996; 46 suppl 2: A294. Hunter SF, Asakura K, Miller DJ, Rodriguez M. The repair of central nervous system myelin: cell biology and pharmacologic action of remyelination-promoting autoantibodies. In: Devon RM, Doucette R, Juurlink BHJ, Nazavali AJ, Schreyer DJ, Verge VMK, editors. Cell biology and pathology of myelin. New York: Plenum Press, 1997: 253–64. Jonas JB, Mardin CY, Schlotzer-Schrehardt U, Naumann GOH. Morphometry of the human lamina cribrosa surface. Invest Ophthalmol Vis Sci 1991; 32: 401–5. Kennedy PGE, Fok-Seang J. Studies on the development, antigenic phenotype; and function of human glial cells in tissue culture. Brain 1986; 109: 1261–77. Kiernan BW, Gotz B, Faissner A, ffrench-Constant C. Tenascin-C inhibits oligodendrocyte precursor cell migration by both adhesiondependent and adhesion-independent mechanisms. Mol Cell Neurosci 1996; 7: 322–35. Kodama T, Hayasaka S, Setogawa T. Myelinated retinal nerve fibers: prevalence, location and effect on visual acuity. Ophthalmologica 1990; 200: 77–83. Minckler TM, Boyd E. Physical growth of the nervous system and its coverings. In: Minkler J, editor. Pathology of the nervous system, Vol. I. New York: McGraw-Hill, NY, 1968: 120–37. Noble M, Murray K. Purified astrocytes promote the in vitro division of a bipotential glial progenitor cell. EMBO J 1984; 3: 2243–7.

Reichenbach A, Schippel K, Schumann R, Hagen E. Ultrastructure of rabbit retinal nerve fibre layer—neuro-glial relationships, myelination, and nerve fibre spectrum. J Hirnforsch 1988; 29: 481–91. Sachsalber A. Schwund markhaltiger NervenFasern in der Netzhaut bei entzuenlicher Atrophie des Sehnerven infolge eines Tumor cerebri. Z Augenheilk 1905; 13: 739–50. Schachat AP, Miller NR. Atrophy of myelinated retinal nerve fibers after acute optic neuropathy. Am J Ophthalmol 1981; 92: 854–6. Scolding NJ, Rayner PJ, Sussman J, Shaw C, Compston DA. A proliferative adult human oligodendrocyte progenitor. Neuroreport 1995; 6: 441–5. Sharpe JA, Sanders MD. Atrophy of myelinated nerve fibres in the retina in optic neuritis. Br J Ophthalmol 1975; 59: 229–32. Small RK, Riddle P, Noble M. Evidence of migration of oligodendrocyte–type-2 astrocyte progenitor cells into the developing rat optic nerve. Nature 1987; 328: 155–7. Straatsma BR, Heckenlively JR, Foos RY, Shahinian JK. Myelinated retinal nerve fibers associated with ipsilateral myopia, amblyopia, and strabismus. Am J Ophthalmol 1979; 88: 506–10. Straatsma BR, Foos RY, Heckenlively JR, Taylor GN. Myelinated retinal nerve fibers. Am J Ophthalmol 1981; 91: 25–38. Targett MP, Sussman J, Scolding N, O’Leary MT, Compston DAS, Blakemore WF. Failure to achieve remyelination of demyelinated rat axons following transplantation of glial cells obtained from the adult human brain. Neuropathol Appl Neurobiol 1996; 22: 199–206.

2082

S. F. Hunter et al.

Tontsch U, Archer DR, Dubois-Dalcq M, Duncan ID. Transplantation of an oligodendrocyte cell line leading to extensive myelination. Proc Natl Acad Sci USA 1994; 91: 11616–20. van der Knaap MS, Valk J. Myelination and retarded myelination. In: van der Knaap MS, Valk J. Magnetic resonance of myelin, myelination, and myelin disorders. 2nd ed. Berlin: Springer, 1995: 31–52. Vrabec F. The temporal raphe of the human retina. Am J Ophthalmol 1966; 62: 926–38.

Wolswijk G, Noble M. Identification of an adult-specific glial progenitor cell. Development 1989; 105: 387–400. Wren D, Wolswijk G, Noble M. In vitro analysis of the origin and maintenance of O-2A adult progenitor cells. J Cell Biol 1992; 116: 167–76. Yakovlev PI, Lecours AR. The myelogenetic cycles of regional maturation of the brain. In: Minkowski A, editor. Regional development of the brain in early life. Oxford: Blackwell, 1967: 3–70.

Webb A, Clark P, Skepper J, Compston A, Wood A. Guidance of oligodendrocytes and their progenitors by substratum topography. J Cell Sci 1995; 108: 2747–60.

Yanoff M, Fine BS. Retina. In: Yanoff M, Fine BS. Ocular pathology: a text and atlas. 3rd ed. Philadelphia: Lippincott, 1989: 381–2.

Weidenheim KM, Bodhireddy SR, Rashbaum WK, Lyman WD. Temporal and spatial expression of major myelin proteins in the human fetal spinal cord during the second trimester. J Neuropathol Exp Neurol 1996; 55: 734–45.

Received April 22, 1997. Revised June 20, 1997. Accepted June 30, 1997