Distinctions between fetal and adult human platelet-derived growth factorЦresponsive neural precursors.код для вставкиСкачать
ORIGINAL ARTICLES Distinctions between Fetal and Adult Human Platelet-Derived Growth Factor– Responsive Neural Precursors Andrew Chojnacki, PhD,1 John J. P. Kelly, MD,1 Walter Hader, MD,2 and Samuel Weiss, PhD1 Objective: Platelet-derived growth factor (PDGF)–responsive neural precursors (PRPs; also known as oligodendrocyte progenitor cells) are one of the best characterized precursor cell populations of the rodent central nervous system. Yet little is known about the biology of human PRPs because of an apparent inability to culture and expand them in large numbers. This study was designed to establish an approach that allows direct comparisons between the biology of fetal and adult human PRPs, as a means to address potential differences in intrinsic myelin-production capabilities. Methods: We used the neurosphere culture system, under low plating density, to isolate, culture, and compare the properties of fetal and adult human PRPs. Results: PDGF stimulated fetal human PRPs to generate neurospheres that differentiated primarily into oligodendrocytes, which acquired myelin basic protein expression, as well as neurons and a small number of astrocytes. Together with PDGF, fibroblast growth factor 2 promoted fetal human PRP expansion. In contrast, adult human PRPs isolated from the corpus callosum required twice the culture period to generate neurospheres, which contained oligodendrocytes, as well as astrocytes, but not neurons. Strikingly, fibroblast growth factor 2 did not promote adult human PRP self-renewal. Interpretation: Differences in the intrinsic proliferation, phenotype, and self-renewal properties of fetal and adult human PRPs suggest they are distinct populations, which may result in distinct myelin-production capabilities. Ann Neurol 2008;64:127–142 Multiple sclerosis (MS) is thought to be an autoimmune disease of the central nervous system that is characterized by inflammatory cell infiltration, astrogliosis, and demyelination.1 Limited remyelination has been observed in chronic silent and active lesions,2 as well as in acute MS lesions.3– 6 Remyelination of the central nervous system in rodents has been demonstrated to require newly generated oligodendrocytes.7 Therefore, an in-depth knowledge of human oligodendrocyte progenitor cell (OPC) biology is crucial for understanding why remyelination in MS ultimately fails. OPCs are one of the best-characterized precursor cell populations of the rodent central nervous system,8 and rodent models of MS, such as experimental autoimmune encephalomyelitis, have been used for the development of drugs to treat the human disease. However, this has sometimes led to the development of strategies that were efficacious in experimental autoimmune encephalomyelitis but were either ineffective or exacer- bated (Lenercept) MS in humans,9 and this may be due, at least in part, to species-specific differences. Little is known about how closely the biology of human OPCs mirrors that of rodent OPCs largely because of difficulties in culturing and expanding either fetal10,11 or adult human OPCs.12,13 Furthermore, the requirement of unique culture conditions for human OPCs makes it difficult to make direct comparisons of their biology with that of rodent OPCs. Differences in the biology of rodent and human cells have been previously demonstrated in the remyelinating capacity of Schwann cells,14,15 the growth factors required for the self-renewal of embryonic stem cells,16,17 and the capacity of fetal11 and adult18 neural stem cells (NSCs) to generate oligodendrocytes. Wilson and colleagues19 have found that fetal human OPCs were largely similar in growth factor responsiveness to their rodent counterparts with the exception of FGF-2 and neurotrophin 3 (NT-3) actions. Fetal and adult OPCs have also been From the Departments of 1Cell Biology and Anatomy and 2Clinical Neurosciences, Hotchkiss Brain Institute, Faculty of Medicine, University of Calgary, Calgary, Alberta, Canada. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.21421 Received Jun 4, 2007, and in revised form Feb 28, 2008. Accepted for publication Apr 4, 2008. This article includes supplementary materials available via the Internet at http://www.interscience.wiley.com/jpages/0364-5134/suppmat Address correspondence to Dr Weiss, Department of Cell Biology and Anatomy, 3330 Hospital Drive, NW, University of Calgary, Faculty of Medicine, Calgary, Alberta, Canada T2N 4N1. E-mail: email@example.com © 2008 American Neurological Association Published by Wiley-Liss, Inc., through Wiley Subscription Services 127 successfully isolated from human brain tissue using fluorescence-activated cell sorting for the OPC marker A2B5.20,21 Importantly, these studies demonstrated differences in the biology of fetal and adult human OPCs that suggest they may have distinct origins. The discovery of two separate OPC populations in the rodent spinal cord22,23 and three separate populations in the forebrain24 supports the contention that fetal and adult human OPCs may be distinct populations. A better understanding of the similarities and differences between rodent and human OPCs, as well as fetal and adult human OPCs, will contribute essential information for understanding remyelination and developing repair strategies for MS. Standardizing cell culture systems used to expand orthologous precursor populations will increase the likelihood that observed differences are meaningful. Using the neurosphere culture system,25 we previously reported the isolation of fetal mouse OPCs that we termed platelet-derived growth factor (PDGF)–responsive precursors (PRPs), because in addition to a robust oligodendroglial differentiation capacity, these precursors also generated neurons.26 We reasoned that if human and rodent PRPs possess similarities, then the same culture protocol utilized for expansion of mouse PRPs should be sufficient to expand PRPs isolated from human brain tissue. Therefore, the neurosphere culture system was used without modification to grow fetal and adult human PRPs. We report that fetal human brain and adult human corpus callosum white matter contain PRPs that generate neurospheres when cultured in the presence of PDGF. Fetal human PRPs demonstrate remarkable conservation of their proliferation, self-renewal, and differentiation properties, when compared with fetal rodent OPCs/PRPs. However, we observed differences in the proliferation and selfrenewal properties of fetal versus adult human PRPs, which may explain, in part, the exhaustion of remyelinating capacity in the adult human brain. Materials and Methods Cell Culture Thirty-two fetal human brain tissue samples, aged 11 to 23 gestational weeks were obtained at therapeutic abortions. The tissue was rinsed three times with phosphate-buffered saline (PBS), with the meninges and other nonneural tissue removed. To assess the ability of different growth factors to induce proliferation using the neurosphere culture system,25 we mechanically dissociated fetal brain tissue with a P1000 pipetter in enzyme-free media hormone mix (MHM) whose composition was described previously.25 The dissociated cell suspension was then cultured at 20,000 to 40,000 cells/ml for 3 weeks, with half the medium being replaced once every 7 days. The formation of neurospheres in epidermal growth factor (EGF; human recombinant; 20ng/ml), FGF-2 (human 128 Annals of Neurology Vol 64 No 2 August 2008 recombinant; 20ng/ml; includes 2g/ml heparan sulfate), PDGF-AA (human recombinant; 100ng/ml), EGF ⫹ FGF-2, EGF ⫹ PDGF-AA, FGF-2 ⫹ PDGF, and EGF ⫹ FGF-2 ⫹ PDGF-AA (all from Peprotech, Rocky Hill, NJ, except for FGF-2, which was from R&D Systems, Minneapolis, MN) was assessed. MHM used to generate neurospheres also contained 2% B27 (Invitrogen, La Jolla, CA). The characteristics of the generated neurospheres were then examined as detailed later. The phenotype potential of the generated neurospheres was examined by washing the neurospheres three times with MHM, and plating them as whole neurospheres or as singlecell suspensions on poly-L-ornithine– and/or laminin-coated glass coverslips. Whole neurospheres were differentiated for 3 days in MHM or in the presence of 20ng/ml leukemia inhibitor factor (LIF; human recombinant; Chemicon, Temecula, CA), 20ng/ml T3 (Sigma, St. Louis, MO), or LIF ⫹ T3. Next, the differentiated neurospheres were assessed for their phenotype potential using indirect immunohistochemistry as detailed later. In addition, neurospheres dissociated to a single-cell suspension were plated at 100,000 cells/ml onto poly-L-ornithine–coated coverslips, differentiated for 3 and 5 days, and assessed for their phenotype potential as detailed later. Self-renewal capacity was assessed by dissociating generated neurospheres and culturing them at 20,000 cells/ml in the factor used for their derivation and various other growth factor combinations. The total numbers of cells generated after 3 weeks of culture were counted. Adult human brain samples taken during temporal lobe resections (6 samples ranging in age from 29 – 40 years; average, 36 years) or corpus callosotomies (5 samples, ranging in age from 8 – 41 years, average, 21 years) were obtained from the Foothills Hospital (Calgary, Alberta, Canada). The tissue was rinsed with PBS, minced with a scalpel into 1mm cubes, and then digested in MHM containing kynurenic acid (0.2mg/ml), DNASE 1 (1mg/ml; Sigma), and purified collagenase (200U/ml; Worthington) for 1.5 hours at 37°C, constantly stirring. The digest was then mechanically dissociated with a P1000 pipetter set at 800l, and rinsed three times with MHM to clear the remaining enzymes. Subsequently, the cell suspension (including the myelin debris) was plated at low (9 cells/mm2) or high (27 cells/mm2) densities into MHM-containing flasks. After 2 hours, the majority of the live cells had attached to the flask bottom, and therefore all the media was removed and replaced to rid the culture of the myelin debris. Cultures were maintained for up to 9 weeks with half the medium being replaced with fresh medium every 7 days. Immunofluorescence Differentiated whole neurospheres or single-cell suspensions of the neurospheres were fixed for 20 minutes in ice-cold 4% paraformaldehyde after the appropriate differentiation period. Coverslips were then processed for immunocytochemistry against mouse IgM anti-O4 (1:50; Chemicon, Temecula, CA), mouse anti-␤-III-tubulin (1:1000; Sigma), rabbit anti–glial fibrillary acidic protein (GFAP; 1:300; BTI, Stoughton, MA), rabbit anti-calretinin (1:200; Swant, Bellinzona, Switzerland), rabbit anti-GABA (1:500; Sigma), goat anti-parvalbumin (1:200: Swant), rabbit anti–human myelin basic protein (MBP; 1:200; Dako, Mississauga, Ontario, Canada), goat anti–human PDGFR␣ (1:10; R&D Systems), biotin-conjugated goat anti–human PDGFR␣ (1:5; R&D Systems), sheep anti–mouse epidermal growth factor receptor (EGFR; 1:50; Biodesign International, Kennebunk, ME), and/or rabbit anti–mouse fibroblast growth factor receptor-2 (FGFR2; 1:50; Santa Cruz Biotechnology). The appropriate biotin-conjugated antibodies were used followed by incubation with streptavidin-Cy3 (1:2,000) for the detection of the mouse IgM anti-O4 and rabbit anti–mouse FGFR2 antibodies. Biotin-conjugated goat anti–human PDGFR␣ was detected by incubation with streptavidin-Cy3 (1:2,000). All other antibodies were detected by incubation with the appropriate rhodamine- or fluorescein isothiocyanate–conjugated donkey secondary antibodies. In addition, Hoechst 33258 (1:100 –1:1,000; Sigma) was used to label nuclei. Fetal human brain tissue was fixed in 4% paraformaldehyde followed by incubation in 10 and 25% sucrose before embedding in Tissue Tek OCT (Sakura Finetek, Torrance, CA). The tissue was then sectioned at 8m and allowed to adhere to the slides overnight. Immunostaining for PDGFR␣ (1:20), EGFR, and FGFR2 was performed as described earlier for the cultured cells. Fetal and adult human brain sections were also processed for staining against PDGFR␣ (1:20) together with rabbit anti-OLIG2 (1:200; gift from Dr Masato Nakafuku), rabbit anti-NG2 (1:20; gift from Dr Bill Stallcup), rabbit anti-S100␤ (1:50; RD, Concord, MA), or mouse antiS100␤ (1:50; Sigma). Slides stained for OLIG2 were pretreated with a biotin blocking kit according to the manufacturer’s instructions (Vectastain, Burlingame, CA). Sections were incubated with the primary antibodies overnight at room temperature in PBS containing 0.3% Triton X-100 (Sigma), followed by three washes with PBS, and a 1-hour block in PBS containing 10% normal donkey serum. For OLIG2 staining, a donkey anti–rabbit biotin-conjugated secondary antibody (1:200) was applied for 1 hour, followed by incubation with streptavidin-fluorescein isothiocyanate (1:500). NG2, both S100␤ antibodies, and PDGFR␣ were detected using the appropriate rhodamine- or fluorescein isothiocyanate–conjugated donkey secondary antibodies. All secondary antibodies were obtained from Jackson ImmunoResearch Laboratories (West Grove, PA). All slides were mounted with Fluorsave (Calbiochem, San Diego, CA). Images were captured on an Axiocam camera (Zeiss, Thornwood, NY) mounted on a Zeiss Axioplan2 microscope (Zeiss) using Axiovision software, and figures were composed in Adobe Photoshop 6.0 (Adobe Systems, Mountain View, CA). Confocal images where captured on a Olympus Optical (Tokyo, Japan) Fluoview BX-50 laser scanning confocal microscope using Fluoview software. Analysis of Self-Renewal Capacity An example is provided demonstrating how the numbers for Figure 4A were generated. For primary culture, the cells were plated at 20,000 cells/ml in 10ml media containing PDGF. Therefore, a total of 200,000 cells were plated. Only about 0.25% of the plated cells generate neurospheres after 3 weeks in culture, whereas the remainder of the cells remained ad- hered to the flask bottom and differentiated or died. Only the neurospheres, which are suspended in the media, were collected. The neurospheres were dissociated, and a cell count was performed. On average, 200,000 cells were generated from these primary PDGF neurospheres. Thus, in addition to the cells that remained adhered to the flask bottom, which either died or differentiated, 200,000 newly generated live cells were produced in the presence of PDGF. This does not include the cells that died during the mechanical dissociation of the neurosphere and is most likely an underestimate. Thus, 200,000 cells were plated, and 200,000 new cells were generated, which results in 1 cell plated generating 1 cell. Statistical Analysis Statistical analysis was conducted using Prism statistical software by GraphPad. Values are mean ⫾ standard error of the mean. For multiple-group comparisons, a one-way analysis of variance was performed followed by a Tukey’s multiplecomparison test. A Student’s t test was performed for comparisons between two groups. p ⬍ 0.05 was considered significant. Results Fetal Human Brain Tissue Contains Platelet-Derived Growth Factor–Responsive Neural Precursors That Are Distinct from Epidermal Growth Factor– and Fibroblast Growth Factor–Responsive Precursors To establish whether fetal human brain tissue contained OPCs similar to those found in rodents, we first examined fetal human brain tissue using immunohistochemistry. We previously demonstrated that fetal mouse PRPs express PDGFR␣ and nuclear-localized FGFR2 in vivo.26 We found that in sections of a whole human brain aged 18 gestational weeks, obtained from the same tissue later used for cell culture experiments, all PDGFR␣-expressing cells colabeled for nuclear FGFR2 (Figs 1A–C). We also found cells that expressed FGFR2 only (presumably FGF-responsive precursors) but not PDGFR␣ (Fig 1B, arrowhead). Examination of PDGFR␣ and FGFR2 expression in dissociated primary cells plated for 24 hours in the absence of growth factors demonstrated that PDGFR␣expressing cells continued to coexpress FGFR2, which was localized either to the nucleus or to the cell membrane (see Figs 1D–G). We found that in contrast with the exclusive nuclear localization of FGFR2 in PDGFR␣-expressing cells in vivo, the majority of the FGFR2 staining was localized to the cell membrane in vitro, which supports previous findings that ligand binding is associated with translocation of FGF receptors to the nucleus.27,28 These data suggested that fetal human brain tissue does contain PRPs, and that they are distinct from FGF-responsive precursors. We have also previously reported that E14 mouse PRPs are distinct from EGF-responsive mouse NSCs.26 Chojnacki et al: Fetal and Adult Human PRPs 129 Fig 1. Fetal human platelet-derived growth factor (PDGF)–responsive precursors are distinct from epidermal growth factor receptor (EGFR)– and fibroblast growth factor receptor 2 (FGFR2)–expressing neural stem cells (NSCs). (A–C) Tissue sections of an 18week-old fetal brain sample immunolabeled for PDGF receptor ␣ (PDGFR␣) (A) and FGFR2 (B). Note that PDGFR␣-positive cells express FGFR2 (arrow) that is localized to the nucleus (C), but not all FGFR2-labeled cells express PDGFR␣ (arrowhead) (B). In fetal brain tissue that had been dissociated, plated in the absence of growth factors, and fixed after 24 hours in culture, some PDGFR␣-positive cells (D) continue to express nuclear-localized FGFR2 (E), but the majority of FGFR2 expression is restricted to the cell membrane (F, G). In fetal brain tissue sections, PDGFR␣ (H) and EGFR (I) labeling were found to be mutually exclusive (J), but PDGFR␣ (K) and EGFR (L) colocalized in culture (K); however, cells expressing only PDGFR␣ (arrowhead) were also observed (M, N). Scale bars ⫽ 10m (A–N); 5m (insets). Examination of fetal human brain sections aged 18 gestational weeks demonstrated that EGFR and PDGFR␣ labeled mutually exclusive populations (see Figs 1H–J). In vitro we found that some cells brightly labeled for 130 Annals of Neurology Vol 64 No 2 August 2008 PDGFR␣ started to express EGFR, and vice versa (see Figs 1K–N). Nevertheless, we still observed PDGFR␣expressing cells that did not label for EGFR (Fig 1M, arrowhead). In addition, sections of a fetal human brain aged 21 gestational weeks also showed that PDGFR␣-expressing cells colabeled for OLIG2, NG2, and S100␤ (see Supplementary Fig 1), all of which have been found to be expressed by OPCs.29 –32 Thus, in vivo, fetal human PRPs represent a distinct population of precursor cells, and this distinction is largely maintained in vitro. Fetal Human Platelet-Derived Growth Factor– Responsive Neural Precursors Generate Neurospheres That Differentiate into Neurons, Oligodendrocytes, and Astrocytes The observation that fetal human brain tissue contained PRPs (see Fig 1) similar to those we found in the fetal mouse brain led us to ask whether we could utilize the neurosphere culture system, previously used to expand fetal mouse PRPs,26 to expand fetal human PRPs. We cultured the dissociated fetal human brain tissue at a low cell density of 20,000 cells/ml because we had previously determined that 95% of the neurospheres generated in PDGF from a 1:1 mix of green fluorescent protein– and non–green fluorescent protein–expressing fetal mouse brain tissue were clonally derived. Fetal human brain tissue, which ranged in age from 11 to 24 gestational weeks, generated neurospheres (50 –200m in size) in PDGF-containing medium after 3 weeks of culture (Fig 2A). PDGF neurospheres were also generated at a density of 10 cells/well for both primary and passaged neurospheres, and also at 1 cell/well for passaged neurospheres (data not shown), indicating that these were, in fact, genuine neurospheres and not the result of cell clumping. Furthermore, neutralizing antibodies directed against PDGFR␣ greatly diminished the generation of both primary and passaged neurospheres in the presence of PDGF (data not shown). PDGF generated more neurospheres compared with medium containing no growth factors ( p ⬍ 0.05; n ⫽ 8), but significantly fewer neurospheres ( p ⬍ 0.001; n ⫽ 8) than all the other conditions tested (see Figs 2B–F). When differentiated for 3 days, PDGF-generated neurospheres differentiated primarily into oligodendrocytes (O4) and neurons (␤-III-tubulin), with only the occasional GFAP-positive astrocyte being observed (see Fig 2G). In contrast with fetal mouse PRPs that generated GABAergic parvalbumin-positive neurons,26 fetal human PRPs generated neurons that were GABAergic and labeled for calretinin but not parvalbumin (data not shown). A larger proportion of EGF-generated progeny differentiated into neurons (13 ⫾ 1%) in comparison with PDGF-generated neurospheres (9 ⫾ 1%; n ⫽ 5; p ⬍ 0.05; see Figs 2G–H). In contrast, PDGF-generated neurospheres contained, on average, at least fivefold as many oligodendrocytes compared with the next best condition (FGF-2 ⫹ PDGF; Figs 3A–F). Furthermore, we found that PDGF stim- Fig 2. Fetal human brain tissue contains precursor cells responsive to platelet-derived growth factor (PDGF) signaling. (A–F) Examples of the neurospheres generated in the presence of the growth factors indicated in the panels from a 21 gestational weeks brain tissue sample after 3 weeks of culture. (G, H) Both PDGF and epidermal growth factor (EGF) neurospheres differentiated for 3 days generated neurons (␤-III-tubulin; green), oligodendrocytes (O4; red), and astrocytes (glial fibrillary acidic protein [GFAP]; blue), but the majority of EGF-generated progeny differentiated into neurons and astrocytes, whereas PDGF-derived progeny differentiated primarily into oligodendrocytes and neurons. Scale bars ⫽ 200m (E); 50m (H). FGF ⫽ fibroblast growth factor; MHM ⫽ media hormone mix. ulation did not enhance the generation of oligodendrocytes by primary EGF ⫹ FGF-2–responsive NSCs (see Fig 3F). It is possible that the inability of PDGF to enhance the generation of oligodendrocytes by primary EGF ⫹ FGF-2–generated neurospheres reflected a limitation of the differentiation conditions used. It has previously Chojnacki et al: Fetal and Adult Human PRPs 131 Fig 3. Fetal human neurospheres generated in the presence of platelet-derived growth factor (PDGF), but in the absence of epidermal growth factor (EGF) or fibroblast growth factor 2 (FGF-2), have the greatest capacity for oligodendroglial differentiation. (A–E) Three-week-old neurospheres generated in various growth factor conditions were differentiated for 3 days in vitro and then stained for the oligodendrocyte-specific antigen O4 (red) and the nuclear stain Hoechst 33258 (blue). (F) A significantly larger proportion of oligodendrocytes (arrows) differentiated from neurospheres cultured in the presence of PDGF, but in the absence of EGF or FGF-2 (A), compared with neurospheres generated in EGF- or FGF-2–containing medium (B–E). *p ⬍ 0.05, PDGF significantly different against all other conditions. Scale bar ⫽ 50m (E). 132 Annals of Neurology Vol 64 No 2 August 2008 been reported that laminin is required for oligodendroglial differentiation from adult human NSCs.18 However, we found that PDGF-generated neurospheres continued to generate significantly more oligodendrocytes than neurospheres generated in the other conditions when differentiated onto coverslips that were coated with laminin together with poly-L-ornithine (see Supplementary Fig 2A). In addition, we tested the capacity of thyroid hormone (T3) and LIF to promote oligodendroglial generation because these factors are known to support oligodendrocyte differentiation33 and survival,34 respectively. We observed no change in the proportion of O4-labeled cells in PDGF neurospheres treated with (see Supplementary Fig 2B) or without T3 (see Fig 3). Almost no oligodendroglial differentiation was observed in LIF-treated neurospheres (see Supplementary Fig 2C). Similarly, oligodendroglial differentiation was almost completely suppressed in neurospheres differentiated in the presence of both LIF and T3 (see Supplementary Fig 2D). Together, the data suggest that PRPs present in the fetal human brain proliferate and generate neurospheres in response to PDGF and have the greatest capacity for oligodendrocyte generation. Fibroblast Growth Factor 2 Promotes the Expansion of Fetal Human Platelet-Derived Growth Factor– Responsive Neural Precursors Although it is well known that the continued expansion of rodent OPCs/PRPs requires the presence of FGF-2 in addition to PDGF,26,35,36 it is unclear whether FGF-2 also promotes the expansion of human OPCs.11,19,20 Thus, we examined the self-renewal properties of human fetal PRPs. In contrast with our findings in the mouse, dissociated primary PDGFgenerated neurospheres could form secondary (pass1) neurospheres in response to PDGF alone (Fig 4A; refer to Materials and Methods for an example of how the numbers were generated). However, the PRP population did not undergo expansion because the number of cells generated equaled the number of cells plated, although this may be an underestimate because some cells are lost during the mechanical dissociation of the neurospheres (see Fig 4C). Fetal mouse PRP expansion and self-renewal,26 as well as the expansion of rodent OPCs, requires the presence of FGF-2 in conjunction with PDGF.35,36 Addition of 20ng/ml FGF-2 to the culture medium increased the size of the secondary neurospheres and prevented their differentiation onto the flask bottom (see Fig 4C). In agreement with the increase in neurosphere size observed, FGF-2 more than doubled the number of cells generated per cells plated (see Fig 4C; p ⬍ 0.05, n ⫽ 5, t test), suggesting that it can promote the expansion of fetal human PRPs. Primary EGF neurospheres derived from the mouse E14 ventral forebrain can generate neurospheres when passaged into PDGF, and these neurospheres are indistinguishable from primary PDGF neurospheres.26 Thus, we tested the potential of EGF ⫹ FGF-2–generated fetal human neurospheres to serve as an alternative source of PDGF neurospheres. We found that primary EGF ⫹ FGF-2 neurospheres could generate pass1 PDGF neurospheres. However, these pass1 PDGF neurospheres (see Fig 4E) derived from primary EGF ⫹ FGF-2 or EGF ⫹ FGF-2 ⫹ PDGF neurospheres generated significantly fewer ( p ⬍ 0.001) oligodendrocytes than primary PDGF neurospheres (see Fig 4D) or pass1 PDGF neurospheres (see Fig 4E) derived from primary PDGF neurospheres. Thus, regardless of the passage, PDGF neurospheres always generated the greatest proportion of oligodendrocytes. It has previously been reported that the capacity of human NSCs to generate oligodendrocytes decreases through subsequent passages.11 In agreement, we found that the capacity for oligodendroglial differentiation decreased with increasing passage, regardless of the growth factors that were used to generate neurospheres (see Fig 4F). Interestingly, by binning the data into early gestational age brain tissue (average, 13 weeks; range, 11–14 weeks) and late gestational age tissue (average, 20 weeks; range, 17–23 weeks), we found that neurospheres generated from younger brain tissue generated significantly more oligodendrocytes per neurosphere (see Fig 4G; p ⬍ 0.001, n ⫽ 7 each condition, t test). Together, the data suggest that although human PRPs can be passaged, their capacity for oligodendroglial differentiation declines with passage, at least in vitro. Fetal Human Platelet-Derived Growth Factor– Responsive Neural Precursors Generate Myelin Basic Protein–Expressing Oligodendrocytes We had previously observed that fetal mouse PRPs generated cells that could express immature neuronal and oligodendroglial antigens simultaneously.26 Therefore, to ensure we were, in fact, identifying oligodendrocytes, we examined the progeny of fetal human PDGF-generated neurospheres for expression of the mature oligodendroglial marker MBP. In addition, we were concerned that oligodendroglial differentiation in larger neurospheres might not be observed because of the limited migration out of the core of neurospheres in some of the conditions. Hence, we cultured neurospheres in PDGF, EGF ⫹ FGF-2, or EGF ⫹ FGF2 ⫹ PDGF for 3 weeks, and then dissociated the neurospheres, plated the suspension at 100,000 cells/ml, and differentiated the cells for 3 or 5 days in vitro. After the appropriate differentiation period, the cultures were dual-labeled for O4 and MBP. Again, we Chojnacki et al: Fetal and Adult Human PRPs 133 Fig 4. Fetal human platelet-derived growth factor (PDGF)–responsive neural precursors generate the greatest proportion of oligodendrocytes over multiple passages. Digital images of 2-week-old secondary neurospheres generated in PDGF (A) or PDGF ⫹ fibroblast growth factor 2 (FGF-2) (B) derived from primary 3-week-old PDGF neurospheres. (C) The capacity of primary cells to proliferate in response to PDGF was compared with progeny passaged into PDGF or PDGF ⫹ FGF-2. PDGF-responsive precursors exhibit significantly greater expansion when passaged into PDGF ⫹ FGF-2–containing medium ( p ⬍ 0.05, t test; n ⫽ 5), in comparison with medium containing PDGF alone. Comparisons of the capacity of primary (D) and pass1 (E) neurospheres to generate oligodendrocytes demonstrate that only neurospheres originally derived in PDGF retain their capacity to generate a large proportion of oligodendrocytes (minimum p ⬍ 0.05, Tukey’s Honest Significant Difference). (F) The capacity to generate oligodendrocytes declines with each subsequent passage regardless of the growth factors used to generate the primary neurospheres (minimum p ⬍ 0.05, Tukey’s Honest Significant Difference). This decrease in oligodendroglial capacity is also seen with PDGF-generated neurospheres derived from older gestational age brain samples (G; *p ⬍ 0.05). Scale bars ⫽ 100m (A, B). found that neurospheres generated in PDGF generated close to fourfold more O4⫹ oligodendrocytes of the live cells plated in comparison with the other conditions, regardless of whether the cells were differentiated for 3 (Figs 5A, B; see Supplementary Fig 3; p ⬍ 0.001; n ⫽ 7) or 5 days in vitro (see Figs 5A, C; see Supplementary Fig 3; p ⬍ 0.001; n ⫽ 5 for P; n ⫽ 6 for E ⫹ F and E ⫹ F ⫹ P). Morphologically, the O4-positive oligodendrocytes generated from PDGF neurospheres after 3 days in vitro were predominantly bipolar, whereas O4-positive cells differentiating from EGF ⫹ FGF-2 ⫹ PDGF neurospheres of the same age were predominantly multipolar and more reminiscent of mature oligodendrocytes. Indeed, a larger proportion of MBP-positive oligodendrocytes (1.4 ⫾ 0.1%; p ⬍ 0.05, n ⫽ 7, Tukey’s HSD) differentiated from 134 Annals of Neurology Vol 64 No 2 August 2008 EGF ⫹ FGF-2 ⫹ PDGF neurospheres after 3 days in vitro than either EGF ⫹ FGF-2– (0.1 ⫾ 0.02) or PDGF-generated (0.7 ⫾ 0.18) progeny (see Fig 5A). However, after 5 days of differentiation, the proportion of MBP-expressing oligodendrocytes was at least 3.5fold greater in PDGF-derived cultures (6.71 ⫾ 0.99%) compared with EGF ⫹ FGF-2 (0.8 ⫾ 0.02%) or EGF ⫹ FGF-2 ⫹ PDGF (1.87 ⫾ 0.21%) cultures ( p ⬍ 0.001; n ⫽ 6 for P; n ⫽ 7 for E⫹F and E⫹F⫹P). Even so, we believe that the numbers of MBP and O4-positive oligodendrocytes we observed in the differentiated dissociates of PDGF-generated neurospheres is an underestimate of their true potential for oligodendroglial differentiation; we observed a large number of oligodendrocytes in this condition without nuclei or with fragmented nuclei that were not included Fig 5. Fetal human platelet-derived growth factor (PDGF)–responsive precursors generate large numbers of myelin basic protein (MBP)–expressing oligodendrocytes. (A) Primary 3-week-old neurospheres generated in epidermal growth factor (EGF) ⫹ fibroblast growth factor receptor 2 (FGF-2), EGF ⫹ FGF-2 ⫹ PDGF, or PDGF were dissociated and differentiated at 100,000 cells/ml for 3 or 5 days in vitro (DIV), and then immunostained for O4 (red), MBP (green), and the nuclear stain Hoechst 33258 (blue). At 3 DIV, the majority of the O4-expressing cells derived from PDGF neurospheres possessed a bipolar morphology reminiscent of immature oligodendrocytes, and few of these cells expressed MBP (B). However, even though EGF ⫹ FGF-2 ⫹ PDGF-generated neurospheres generated significantly less O4⫹ (black bars) oligodendrocytes, a larger proportion of these cells expressed MBP (white bars), and morphologically these cells resembled more mature oligodendrocytes. At 5 DIV, the majority of the O4⫹ cells from PDGF-generated neurospheres had a complex multidendritic morphology, and a significantly larger proportion of these cells coexpressed MBP expression compared with EGF ⫹ FGF-2- or EGF ⫹ FGF-2 ⫹ PDGF-generated neurospheres (C). Scale bar ⫽ 50m Chojnacki et al: Fetal and Adult Human PRPs 135 in the counts (data not shown). Together, the data suggest that fetal human PRPs have the greatest capacity for generating MBP-expressing oligodendrocytes. Platelet-Derived Growth Factor and Neurotrophin 3 Induce the Generation of Neurospheres from the Adult Human Corpus Callosum Remyelination within MS patients has been demonstrated to occur in cases of MS.2,4 – 6 Currently, however, there is little evidence for proliferating OPCs within the adult human brain, and attempts to culture adult human OPCs have been largely unsuccessful. Because we had established the presence of PRPs within fetal brains aged 11 to 24 gestational weeks, which possessed a robust potential for oligodendroglial generation, we asked whether adult human white matter also contained such PRPs, and how similar they were to their fetal counterparts. We obtained samples of white matter from four corpus callosotomies, which included the anterior body of the corpus callosum. For two of the specimens, samples were reserved for immunohistochemistry. We stained 20- and 16-year-old samples for PDGFR␣ and FGFR2 to establish the presence of PRPs in the tissue obtained. We found that 3.6 ⫾ 0.3% (36,713 cells examined, 29,203 from the 20year-old corpus callosum and 7,510 from the 16-yearold corpus callosum) of the cells expressed PDGFR␣ (Fig 6A), and all PDGFR␣ ⫹ cells labeled for nuclearlocalized FGFR2 (see Figs 6A, B), as we had found for fetal human PRPs. However, not all FGFR2-labeled cells expressed PDGFR␣. In addition, we stained the tissue for Ki67, a marker associated with cell proliferation, and PDGFR␣ (see Fig 6C) and found that approximately 0.5% of PDGFR␣-expressing cells were proliferating. We also found that PDGFR␣-expressing cells colabeled for NG2, OLIG2, and S100␤, all of which have been found to be expressed by OPCs29 –32 (see Supplementary Fig 4). Thus, adult human corpus callosum contains PRPs that may contribute to remyelination. Next, we examined whether neurospheres could be generated from the white matter of the corpus callosum. We included NT-3 in the culture medium in conjunction with PDGF, because NT-3 has been reported previously to increase fetal human OPC proliferation.19 Neurospheres that ranged in size from 50 to 200m were generated after 7 weeks in culture in the presence of PDGF ⫹ NT-3 (see Fig 6D). When differentiated for 3 days in vitro, extensive migration was observed away from the center of the neurospheres (see Fig 6E). The cells migrating from the neurospheres had a bipolar morphology with a few smaller processes radiating away from the cell soma (see Fig 6F), characteristic of immature oligodendrocytes and OPCs. Surprisingly, we found that neurospheres were gener- 136 Annals of Neurology Vol 64 No 2 August 2008 ated at only low-density conditions for all samples (see Supplementary Fig 5A), whereas no neurospheres were generated at high cell densities (see Supplementary Fig 5B). We also examined whether we could generate neurospheres in the presence of PDGF and NT-3 from the white matter obtained from temporal lobe resections. Despite the fact that we observed PDGFR␣positive cells in samples of temporal lobe myelin (data not shown), none of the samples generated neurospheres even after 9 weeks in culture (n ⫽ 6). The addition of FGF-2 in combination with PDGF and NT-3 to some of the cultures (n ⫽ 4) also failed to stimulate neurosphere formation. Furthermore, lowdensity cultures of temporal lobe white matter did not promote the generation of neurospheres by PDGF and NT-3. The data suggest that adult human corpus callosum contains PRPs that can be cultured as neurospheres in the presence of PDGF and NT-3. Restricted Phenotypes and Limited Self-Renewal of Adult Human Corpus Callosum–Derived PlateletDerived Growth Factor–Responsive Neural Precursors Rodent postnatal and adult OPCs differ in their proliferation and differentiation characteristics.37 We sought to determine whether these differences existed between fetal and adult human PRPs. Therefore, we first examined the phenotype potential of neurospheres generated in PDGF ⫹ NT-3 from the adult corpus callosum. We found O4-expressing oligodendrocytes and GFAP-expressing astrocytes migrating away from the neurospheres (see Figs 6G, H). After 3 days of differentiation, 33 ⫾ 3% of the cells adopted O4 expression, whereas 15 ⫾ 3% of the cells expressed GFAP (n ⫽ 3). However, in contrast with the fetal human PRPs examined, the majority of the neurospheres did not contain ␤-III-tubulin–expressing neurons (only one neuron observed of all the neurospheres examined; see Figs 6G, H). Furthermore, we occasionally found MBP-expressing cells within the neurospheres after 5 days in vitro (see Fig 6H), suggesting that the progeny of the adult human PRPs have the potential to generate the constituents of myelin. Lastly, we examined the capacity of adult human PRPs derived from the corpus callosum to self-renew. Approximately 10 neurospheres, which were at least 150m in size, were dissociated per 1ml medium. The dissociate was then plated in the presence of PDGF and NT-3 or PDGF, NT-3, and FGF-2. We did not observe the generation of secondary neurospheres even after 12 weeks of culture, regardless of the presence of FGF-2 (n ⫽ 3, data not shown). We did find some evidence of limited proliferation in the form of adherent colonies of cells comprising 10 to 20 cells, and these cells had an astroglial morphology. These data suggest that in contrast with fetal human PRPs, adult Fig 6. Platelet-derived growth factor–responsive neural precursors (PRPs) isolated from adult human corpus callosum–generated neurospheres in response to platelet-derived growth factor (PDGF) ⫹ neurotrophin 3 (NT-3) that differentiated into oligodendrocytes and astrocytes. (A) Section of a 20-year-old human corpus callosum sample stained for PDGF receptor ␣ (PDGFR␣) (red) and fibroblast growth factor receptor 2 (FGFR2) (green) demonstrates that they label the same cell. (B) Confocal microscopy confirms that FGFR2 labels the nucleus of PDGFR␣-expressing cells. Approximately 0.5% of PDGFR␣-expressing cells also labeled for the nuclear antigen Ki67 (C), which is associated with cell proliferation. (D) Examples of the neurospheres generated in PDGF ⫹ NT-3 after 7 weeks and 5 days of culture. (E) Large numbers of cells migrated away from the center of the neurosphere after 3 days of differentiation, and these cells had a bipolar or a polydendritic morphology (F). (G, H) PDGF-generated neurospheres differentiated into oligodendrocytes and astrocytes after 3 days, but neurons were rarely observed. (I) MBP-positive cells were also observed after 5 days of differentiation. Scale bars ⫽ 12.5m (A), 5m (B, C), 100m (D), 25m (E, F), 50m (G), 20m (H), 10m (I). GFAP ⫽ glial fibrillary acidic protein. human corpus callosum PRPs are limited in their phenotype and self-renewal potentials. Discussion Utilizing the same culture system we previously used to examine the properties of fetal mouse forebrain PRPs,26 we have cultured and examined the properties of fetal and adult human PRPs. We found that fetal human PRPs, like fetal mouse PRPs, were distinct from EGF- and FGF-responsive NSCs, and generated the greatest numbers of oligodendrocytes. Fetal human PRPs self-renewed poorly in PDGF, but expanded in the presence of FGF signaling, and continued to generate the greatest proportion of oligodendrocytes after passaging. In addition, we found that adult human PRPs isolated from the corpus callosum, but not the temporal lobe, could generate neurospheres in response to PDGF and NT-3, and had a robust capacity for oligodendroglial generation. However, adult human PRPs were distinct from fetal PRPs by the prolonged Chojnacki et al: Fetal and Adult Human PRPs 137 culture period required for the generation of neurospheres, the inability of FGF-2 to promote their selfrenewal, and their inability to generate neurons or a large number of MBP-expressing oligodendrocytes. Properties of Mouse and Human Platelet-Derived Growth Factor–Responsive Neural Precursors/ Oligodendrocyte Progenitor Cells Are Remarkably Similar One of the aims of this study was to determine the degree to which the biology of rodent OPCs is a predictor of human OPC biology (Table). Use of the same in vitro system to culture both mouse and human precursor populations enabled the pursuit of this aim. Many studies have reported that PDGF failed to stimulate the proliferation of human OPCs11–13,19 despite its well-known capacity to stimulate rodent PRP/OPC proliferation.26,38,39 We observed that PDGF induced proliferation and the generation of neurospheres by fetal human PRPs, just as it did for fetal mouse forebrain PRPs.26 It is possible that the differences observed in PDGF responsiveness simply stems from the different culture systems used. Dissociating the tissue mechanically, which we found to be quite effective, without utilizing enzymes may have increased responsiveness of our cell preparations to growth factors as we have found that enzymatic digestion can adversely affect cell proliferation in response to mitogens (unpublished observations). Nevertheless, Wilson and colleagues19 have also found that PDGF can stimulate the proliferation of fetal human OPCs, albeit those derived from the spinal cord. We did, however, find some differences with respect to the effect of PDGF on the formation of neurospheres by fetal human PRPs. We found that it took three times the culture period for fetal human PRPs to generate neurospheres of the same size compared with fetal mouse PRPs. We also found that in contrast with mouse PRPs, human PRPs passaged in the presence of PDGF. Whether these discrepancies are a result of an increased cell cycle time or differences in survival/apoptosis requires further investigation. Thus, although both fetal rodent and human PRPs/ OPCs proliferate in response to PDGF, subtle differences in PDGF responsiveness do exist between the two populations. In addition to similarities in growth factor responsiveness, we found that mouse and fetal human PRPs were similar in their phenotype potentials (see the Table). Both generate a large proportion of oligodendrocytes that can differentiate into a significant number of MBP-expressing oligodendrocytes, and both have the capacity to generate neurons. However, fetal human PRPs generated calretinin-positive interneurons in contrast with the parvalbumin-positive interneurons generated by fetal mouse PRPs.26 Both precursor populations also have the capacity to generate astrocytes. Nonetheless, fetal human PRPs appear to generate astrocytes more readily because they were observed even in control medium, whereas the generation of astrocytes by fetal mouse PRPs required the presence of bone morphogenetic protein-2 and ciliary neurotrophic factor.26 We are uncertain of the significance of these Table. Characteristics of Mouse and Human Platelet-Derived Growth Factor–Responsive Neural Precursors Characteristics Embryonic PRPs Mouse PDGF stimulates proliferation Time to neurosphere formation, wk Self-renewing in factor of derivation PDGFR␣-expressing cells express nuclear FGFR2 FGF-2 promotes expansion Generate large numbers of oligodendrocytes Generate large numbers of MBP-positive oligodendrocytes Generate neurons Neuronal phenotype Capacity to generate astrocytes Adult PRPs Human Human Yes 1 No Yes Yes 3 Yes Yes Yesa 6-7 No Yes Yes Yes Yes Yes Yes Yes No Yes No Yes GABAergic parvalbuminpositive Yesb Yes GABAergic calretininpositive Yes No N/A a Yes With neurotrophin 3. Requires bone morphogenetic protein-2 and ciliary neurotrophic factor stimulation. PRP ⫽ platelet-derived growth factor–responsive neural precursors; PDGF ⫽ platelet-derived growth factor; PDGFR␣ ⫽ plateletderived growth factor receptor ␣; FGFR2 ⫽ fibroblast growth factor receptor 2; FGF-2 ⫽ fibroblast growth factor 2; MBP ⫽ myelin basic protein. b 138 Annals of Neurology Vol 64 No 2 August 2008 subtle differences, but they highlight the import of verifying results obtained from rodent studies with human studies. FGF-2 is known to promote the self-renewal of rodent optic nerve–derived OPCs35,36,40; however, previous studies of human OPCs have failed to demonstrate any actions of FGF-2 on self-renewal.11,19 We found that both fetal mouse26 and human PRPs express nuclear-localized FGFR2 (see Figs 1A–C), indicative of ligand activation.27,28,41 This suggests that FGFR2 signaling is active in PRPs, and that it may function in promoting their self-renewal. In contrast, Bansal and colleagues42 have found that only mature oligodendrocytes express FGFR2. It is unclear why this discrepancy in FGFR2 expression is observed, but others43,44 have found that FGFR2 messenger RNA is expressed in white matter tracts, which are known to be populated by OPCs in addition to mature oligodendrocytes.37,45 Nevertheless, we observed that FGF-2 promoted the expansion of fetal human PRPs (see Figs 4A–C), just as it did for fetal mouse PRPs. This dissimilarity in the actions of FGF signaling observed by different groups may indicate regional distinctions in OPC growth factor requirements for proliferation/self-renewal, given that both Wilson and colleagues19 and Chandran and colleagues11 cultured dissociated fetal human spinal cord, whereas we utilized fetal human brain. Using indirect immunohistochemistry, we also demonstrated that fetal human PRPs are distinct from both FGF- and EGF-responsive fetal human NSCs/precursors (see Fig 1). In agreement with previous findings,11,46,47 we found that fetal human NSCs were severely restricted in their capacity to generate oligodendrocytes (less than 4% of the cells generated) regardless of the differentiation condition (see Figs 3, 4, 5, and Supplementary Fig 2). Also, we could not generate PDGF neurospheres with the same oligedendroglial capacity from EGF ⫹ FGF-2 expanded human NSCs, despite the fact that fetal mouse NSCs possess this capability.26 It is possible that some additional factors may be required for the generation of PRPs and/or oligodendrocytes from fetal human NSCs. In summary, although they differ in some respects, the biology of fetal mouse PRPs is, in principle, predictive of fetal human PRP biology. Oligodendroglial Differentiation Capacity of Fetal Human Platelet-Derived Growth Factor–Responsive Neural Precursors Declines with Passage and Gestational Age Our results demonstrate that the capacity of both fetal human NSCs and PRPs to generate oligodendrocytes decreases through subsequent passages. This is in agreement with the findings of Chandran and colleagues,11 who observed that neurospheres generated in FGF-2 from fetal human spinal cord had a limited capacity for generating GalC-positive oligodendrocytes, and that later passage neurospheres generated even fewer oligodendrocytes. Bone morphogenetic proteins, previously shown to inhibit oligodendroglial differentiation in rodents,48 –51 may contribute to the regulation of oligodendrocyte differentiation by fetal human PRPs. Indeed, adult human OPCs, purified using magnetic activated cell sorting for A2B5, express bone morphogenetic protein-2 and -7.52 The receptor tyrosine phosphatase ␤/ may also participate in regulating the differentiation of oligodendrocytes generated by fetal human PRPs because Sim and colleagues52 found that it regulates the differentiation of adult human OPCs. Importantly, we found that the potential of PRPs to generate oligodendrocytes decreased with an increase in the age of the fetal brain tissue (see Fig 4G), suggesting that this restriction also occurs in vivo. Whether adult human PRPs also demonstrate a similar decrease in the capacity to generate oligodendrocytes through subsequent divisions needs to be tested because this may also contribute to the limited remyelination observed in MS. Fetal and Adult Human Platelet-Derived Growth Factor–Responsive Neural Precursors Are Distinct Populations Similar to fetal human PRPs, adult human corpus callosum–derived PRPs proliferated and generated neurospheres in the presence of PDGF, and generated large numbers of oligodendrocytes. However, we observed several differences between fetal and adult human PRPs. Seven weeks, compared with 3 weeks for fetal human PRPs, were required for the generation of neurospheres of comparable size from adult human brain tissue. Also, we rarely observed MBP-expressing cells after 5 days in vitro from adult human PRPs, whereas upward of 7% of fetal human PRP progeny expressed MBP after 5 days in vitro. These findings are similar to the increased cell-cycle time of adult rodent OPCs, and the lengthened period required for oligodendroglial maturation of adult compared with postnatal rodent OPCs.37 However, the adult brain tissue was obtained from epilepsy patients, and how this may have influenced the biology of adult PRPs is unknown. Also, in contrast with the neurogenic capacity of fetal human PRPs, adult human corpus callosum PRPs essentially failed to generate neurons, which is similar to our findings for PRPs obtained from the adult mouse corpus callosum53 and adult O-2A progenitor cells.45 The differences that fetal and postnatal PRPs/OPCs have with their adult counterparts, whether rodent or human, serve to highlight the similarities between adult rodent and human PRPs/OPCs. We did, however, find an important difference in the self-renewal/expansion capacity of adult human Chojnacki et al: Fetal and Adult Human PRPs 139 PRPs compared with adult rodent OPCs, as well as fetal human PRPs. This was despite the fact that fetal PRPs were cultured in the presence of PDGF to generate neurospheres, whereas adult PRPs were cultured in the presence of PDGF ⫹ NT-3, which could not account for the differences we observed because fetal PRPs cultured in PDGF ⫹ NT-3 maintain their capacity to self-renew and their phenotype potential (data not shown). In contrast with its actions on fetal human PRPs and adult rodent OPCs,54 we found that FGF-2 did not promote the expansion/proliferation of adult PRPs isolated from the corpus callosum (see the Table). Previous studies have demonstrated that unknown factors secreted by cortical astrocytes can abrogate the effects of FGF signaling on adult rodent optic nerve OPCs.55 However, it is unlikely that autocrine/paracrine effects reduced the ability of FGF-2 to promote self-renewal of adult human PRPs in our cultures, given the low cell culture density used. It is possible that other factors may contribute to the maintenance of adult human PRP proliferation. Moreover, it is tempting to speculate that the inability of FGF signaling to promote the expansion of adult human PRPs may be a factor contributing to the limited remyelination observed in human demyelinating diseases. Whether adult OPCs continue to display limited selfrenewal in the presence of inflammation, denuded axons, or myelin inhibitory factors remains to be determined. Taken together, the results strongly suggest that fetal and adult human PRPs are distinct populations of precursors. In support of such a conclusion, it has been reported that adult human OPCs/PRPs ensheathed more host axons per donor cell and generated oligodendrocytes more efficiently than their fetal counterparts when transplanted into shiverer mice.21 Furthermore, evidence from rodent studies has demonstrated at least three different OPC populations in the mouse forebrain, in which the earliest population is largely replaced by the later two.24 Finally, several lines of evidence also suggest that adult human corpus callosum–derived PRPs and temporal lobe subcortical white matter OPCs may be distinct precursor populations. We were unable to generate neurospheres from white matter obtained from human temporal lobe resections. This is in contrast with the findings of Nunes and colleagues,56 because they were able to generate neurospheres from dissociates of the human temporal lobe tissue that had been sorted for the A2B5 antigen. They also found that only 3 weeks was required for the generation of neurospheres from the adult temporal lobe, in contrast with the 7 weeks we found are needed for the adult human corpus callosum. Moreover, the observations that corpus callosum–derived PRPs, in contrast with temporal lobe OPCs,20,56 do not readily generate neurons or 140 Annals of Neurology Vol 64 No 2 August 2008 self-renew in the presence of FGF-2 further supports the conclusion that the two populations are distinct. Again, whether this is due to culture system differences or distinct precursor populations is unknown. However, the observations that focal MS lesions found in the human brain predominantly localize to the periventricular white matter of the frontal and parietal lobes,57–59 whereas only 30% of MS patients have focal lesions in the corpus callosum,60 may be indicative of underlying differences in oligodendrocyte populations and/or their precursors in specific regions of the adult human brain. This work was supported by the Neuroscience Canada Foundation, the Multiple Sclerosis Research Foundation of Canada, and the Alberta Heritage Foundation for Medical Research (J.J.P.K., S.W.). We thank Drs M. Noble, M. Raff, and V. W. Yong for reviewing a previous version of this manuscript. We also thank R. Hassam and D. Livingstone for excellent technical assistance. References 1. Frohman EM, Racke MK, Raine CS. Multiple sclerosis—the plaque and its pathogenesis. N Engl J Med 2006;354:942–955. 2. Raine CS, Scheinberg L, Waltz JM. Multiple sclerosis. Oligodendrocyte survival and proliferation in an active established lesion. Lab Invest 1981;45:534 –546. 3. Wu E, Raine CS. Multiple sclerosis. Interactions between oligodendrocytes and hypertrophic astrocytes and their occurrence in other, nondemyelinating conditions. Lab Invest 1992;67: 88 –99. 4. Prineas JW, Kwon EE, Goldenberg PZ, et al. Multiple sclerosis. Oligodendrocyte proliferation and differentiation in fresh lesions. Lab Invest 1989;61:489 –503. 5. Prineas JW, Kwon EE, Goldenberg PZ, et al. Interaction of astrocytes and newly formed oligodendrocytes in resolving multiple sclerosis lesions. Lab Invest 1990;63:624 – 636. 6. Raine CS, Wu E. Multiple sclerosis: remyelination in acute lesions. J Neuropathol Exp Neurol 1993;52:199 –204. 7. Keirstead HS, Blakemore WF. The role of oligodendrocytes and oligodendrocyte progenitors in CNS remyelination. Adv Exp Med Biol 1999;468:183–197. 8. Richardson WD, Kessaris N, Pringle N. Oligodendrocyte wars. Nat Rev Neurosci 2006;7:11–18. 9. TNF neutralization in MS: results of a randomized, placebocontrolled multicenter study. The Lenercept Multiple Sclerosis Study Group and the University of British Columbia MS/MRI Analysis Group. Neurology 1999;53:457– 465. 10. Zhang SC, Ge B, Duncan ID. Tracing human oligodendroglial development in vitro. J Neurosci Res 2000;59:421– 429. 11. Chandran S, Compston A, Jauniaux E, et al. Differential generation of oligodendrocytes from human and rodent embryonic spinal cord neural precursors. Glia 2004;47:314 –324. 12. Scolding NJ, Rayner PJ, Sussman J, et al. A proliferative adult human oligodendrocyte progenitor. Neuroreport 1995;6: 441– 445. 13. Armstrong RC, Dorn HH, Kufta CV, et al. Pre-oligodendrocytes from adult human CNS. J Neurosci 1992;12:1538 –1547. 14. Morrissey TK, Bunge RP, Kleitman N. Human Schwann cells in vitro. I. Failure to differentiate and support neuronal health under co-culture conditions that promote full function of rodent cells. J Neurobiol 1995;28:171–189. 15. Morrissey TK, Kleitman N, Bunge RP. Human Schwann cells in vitro. II. Myelination of sensory axons following extensive purification and heregulin-induced expansion. J Neurobiol 1995;28:190 –201. 16. Rao M. Conserved and divergent paths that regulate selfrenewal in mouse and human embryonic stem cells. Dev Biol 2004;275:269 –286. 17. Xu RH, Peck RM, Li DS, et al. Basic FGF and suppression of BMP signaling sustain undifferentiated proliferation of human ES cells. Nat Methods 2005;2:185–190. 18. Arsenijevic Y, Villemure JG, Brunet JF, et al. Isolation of multipotent neural precursors residing in the cortex of the adult human brain. Exp Neurol 2001;170:48 – 62. 19. Wilson HC, Onischke C, Raine CS. Human oligodendrocyte precursor cells in vitro: phenotypic analysis and differential response to growth factors. Glia 2003;44:153–165. 20. Roy NS, Wang S, Harrison-Restelli C, et al. Identification, isolation, and promoter-defined separation of mitotic oligodendrocyte progenitor cells from the adult human subcortical white matter. J Neurosci 1999;19:9986 –9995. 21. Windrem MS, Nunes MC, Rashbaum WK, et al. Fetal and adult human oligodendrocyte progenitor cell isolates myelinate the congenitally dysmyelinated brain. Nat Med 2004;10:93–97. 22. Vallstedt A, Klos JM, Ericson J. Multiple dorsoventral origins of oligodendrocyte generation in the spinal cord and hindbrain. Neuron 2005;45:55– 67. 23. Cai J, Qi Y, Hu X, et al. Generation of oligodendrocyte precursor cells from mouse dorsal spinal cord independent of Nkx6 regulation and Shh signaling. Neuron 2005;45:41–53. 24. Kessaris N, Fogarty M, Iannarelli P, et al. Competing waves of oligodendrocytes in the forebrain and postnatal elimination of an embryonic lineage. Nat Neurosci 2006;9:173–179. 25. Reynolds BA, Weiss S. Clonal and population analyses demonstrate that an EGF-responsive mammalian embryonic CNS precursor is a stem cell. Dev Biol 1996;175:1–13. 26. Chojnacki A, Weiss S. Isolation of a novel platelet-derived growth factor-responsive precursor from the embryonic ventral forebrain. J Neurosci 2004;24:10888 –10899. 27. Schmahl J, Kim Y, Colvin JS, et al. Fgf9 induces proliferation and nuclear localization of FGFR2 in Sertoli precursors during male sex determination. Development 2004;131:3627–3636. 28. Wells A, Marti U. Signalling shortcuts: cell-surface receptors in the nucleus? Nat Rev Mol Cell Biol 2002;3:697–702. 29. Deloulme JC, Raponi E, Gentil BJ, et al. Nuclear expression of S100B in oligodendrocyte progenitor cells correlates with differentiation toward the oligodendroglial lineage and modulates oligodendrocytes maturation. Mol Cell Neurosci 2004;27: 453– 465. 30. Nishiyama A, Lin XH, Giese N, et al. Co-localization of NG2 proteoglycan and PDGF alpha-receptor on O2A progenitor cells in the developing rat brain. J Neurosci Res 1996;43: 299 –314. 31. Zhou Q, Wang S, Anderson DJ. Identification of a novel family of oligodendrocyte lineage-specific basic helix-loop-helix transcription factors. Neuron 2000;25:331–343. 32. Lu QR, Yuk D, Alberta JA, et al. Sonic hedgehog–regulated oligodendrocyte lineage genes encoding bHLH proteins in the mammalian central nervous system. Neuron 2000;25:317–329. 33. Ahlgren SC, Wallace H, Bishop J, et al. Effects of thyroid hormone on embryonic oligodendrocyte precursor cell development in vivo and in vitro. Mol Cell Neurosci 1997;9:420 – 432. 34. Barres BA, Schmid R, Sendnter M, Raff MC. Multiple extracellular signals are required for long-term oligodendrocyte survival. Development 1993;118:283–295. 35. McKinnon RD, Matsui T, Dubois-Dalcq M, Aaronson SA. FGF modulates the PDGF-driven pathway of oligodendrocyte development. Neuron 1990;5:603– 614. 36. Bogler O, Wren D, Barnett SC, et al. Cooperation between two growth factors promotes extended self-renewal and inhibits differentiation of oligodendrocyte-type-2 astrocyte (O-2A) progenitor cells. Proc Natl Acad Sci U S A 1990;87:6368 – 6372. 37. Wolswijk G, Noble M. Identification of an adult-specific glial progenitor cell. Development 1989;105:387– 400. 38. Raff MC, Lillien LE, Richardson WD, et al. Platelet-derived growth factor from astrocytes drives the clock that times oligodendrocyte development in culture. Nature 1988;333:562–565. 39. Pringle N, Collarini EJ, Mosley MJ, et al. PDGF A chain homodimers drive proliferation of bipotential (O-2A) glial progenitor cells in the developing rat optic nerve. Embo J 1989;8: 1049 –1056. 40. Gard AL, Pfeiffer SE. Glial cell mitogens bFGF and PDGF differentially regulate development of O4⫹GalC⫺ oligodendrocyte progenitors. Dev Biol 1993;159:618 – 630. 41. Wiedlocha A, Sorensen V. Signaling, internalization, and intracellular activity of fibroblast growth factor. Curr Top Microbiol Immunol 2004;286:45–79. 42. Bansal R, Kumar M, Murray K, et al. Regulation of FGF receptors in the oligodendrocyte lineage. Mol Cell Neurosci 1996;7:263–275. 43. Asai T, Wanaka A, Kato H, et al. Differential expression of two members of FGF receptor gene family, FGFR-1 and FGFR-2 mRNA, in the adult rat central nervous system. Brain Res Mol Brain Res 1993;17:174 –178. 44. Belluardo N, Wu G, Mudo G, et al. Comparative localization of fibroblast growth factor receptor-1, -2, and -3 mRNAs in the rat brain: in situ hybridization analysis. J Comp Neurol 1997; 379:226 –246. 45. Ffrench-Constant C, Raff MC. Proliferating bipotential glial progenitor cells in adult rat optic nerve. Nature 1986;319: 499 –502. 46. Murray K, Dubois-Dalcq M. Emergence of oligodendrocytes from human neural spheres. J Neurosci Res 1997;50:146 –156. 47. Quinn SM, Walters WM, Vescovi AL, Whittemore SR. Lineage restriction of neuroepithelial precursor cells from fetal human spinal cord. J Neurosci Res 1999;57:590 – 602. 48. Zhu G, Mehler MF, Mabie PC, Kessler JA. Developmental changes in progenitor cell responsiveness to cytokines. J Neurosci Res 1999;56:131–145. 49. Yung SY, Gokhan S, Jurcsak J, et al. Differential modulation of BMP signaling promotes the elaboration of cerebral cortical GABAergic neurons or oligodendrocytes from a common sonic hedgehog-responsive ventral forebrain progenitor species. Proc Natl Acad Sci U S A 2002;99:16273–16278. 50. Mabie PC, Mehler MF, Marmur R, et al. Bone morphogenetic proteins induce astroglial differentiation of oligodendroglialastroglial progenitor cells. J Neurosci 1997;17:4112– 4120. 51. Grinspan JB, Edell E, Carpio DF, et al. Stage-specific effects of bone morphogenetic proteins on the oligodendrocyte lineage. J Neurobiol 2000;43:1–17. 52. Sim FJ, Lang JK, Waldau B, et al. Complementary patterns of gene expression by human oligodendrocyte progenitors and their environment predict determinants of progenitor maintenance and differentiation. Ann Neurol 2006;59:763–779. 53. Gregg C, Shikar V, Larsen P, et al. White matter plasticity and enhanced remyelination in the maternal CNS. J Neurosci 2007; 27:1812–1823. Chojnacki et al: Fetal and Adult Human PRPs 141 54. Wolswijk G, Noble M. Cooperation between PDGF and FGF converts slowly dividing O-2Aadult progenitor cells to rapidly dividing cells with characteristics of O-2Aperinatal progenitor cells. J Cell Biol 1992;118:889 –900. 55. Mayer M, Bogler O, Noble M. The inhibition of oligodendrocytic differentiation of O-2A progenitors caused by basic fibroblast growth factor is overridden by astrocytes. Glia 1993;8: 12–19. 56. Nunes MC, Roy NS, Keyoung HM, et al. Identification and isolation of multipotential neural progenitor cells from the subcortical white matter of the adult human brain. Nat Med 2003; 9:439 – 447. 142 Annals of Neurology Vol 64 No 2 August 2008 57. Lazeron RH, Boringa JB, Schouten M, et al. Brain atrophy and lesion load as explaining parameters for cognitive impairment in multiple sclerosis. Mult Scler 2005;11:524 –531. 58. Swirsky-Sacchetti T, Mitchell DR, Seward J, et al. Neuropsychological and structural brain lesions in multiple sclerosis: a regional analysis. Neurology 1992;42:1291–1295. 59. Sperling RA, Guttmann CR, Hohol MJ, et al. Regional magnetic resonance imaging lesion burden and cognitive function in multiple sclerosis: a longitudinal study. Arch Neurol 2001;58:115–121. 60. Simon JH, Holtas SL, Schiffer RB, et al. Corpus callosum and subcallosal-periventricular lesions in multiple sclerosis: detection with MR. Radiology 1986;160:363–367.