GLIA 53:294–303 (2006)

Oligodendrocyte Precursor Cells Generate Pituicytes In Vivo During Neurohypophysis Development ISABELLE VIRARD,1 DELPHINE COQUILLAT,1 MIRCEA BANCILA,1 SOVANN KAING,2 AND PASCALE DURBEC1* 1 Laboratoire de Neurogene`se et Morphogene`se dans le Developpement et chez l’Adulte, CNRS UMR 6156, Universite de la Mediterranee, IBDM, Parc Scientifique de Luminy, Marseille, France 2 Department of Pediatric Oncology, Dana-Farber Cancer Institute, Boston, Massachusetts

KEY WORDS oligodendrocyte precursor cell; pituicyte; neurohypophysis; cell lineage; cell differentiation

ABSTRACT In the vertebrate brain, much remains to be understood concerning the origin of glial cell diversity and the potential lineage relationships between the various types of glia. Besides astrocytes and myelin-forming oligodendrocytes, other macroglial cell populations are found in discrete areas of the central nervous system (CNS). They share functional features with astrocytes and oligodendrocytes but also display specific characteristics. Such specialized cells, called pituicytes, are located in the neurohypophysis (NH). Our work focuses on the lineage of the pituicytes during rodent development. First, we show that cells identified with a combination of oligodendrocyte precursor cell (OPC) markers are present in the developing rat NH. In culture, neonatal NH progenitors also share major functional characteristics with OPCs, being both migratory and bipotential, i.e. able to give rise to type 2 astrocytes and oligodendrocytes. We then observe that, either in vitro or after transplantation into myelin-deficient Shiverer brain, pieces of NH generate myelinating oligodendrocytes, confirming the oligodendrogenic potentiality of NH cells. However, no mature oligodendrocyte can be found in the NH. This led us to hypothesize that the OPCs present in the developing NH might be generating other glial cells, especially the pituicytes. Consistent with this hypothesis, the OPCs appear during NH development before pituicytes differentiate. Finally, we establish a lineage relationship between olig11 cells, most likely OPCs, and the pituicytes by fate-mapping experiments using genetically engineered mice. This constitutes the first demonstration that OPCs generate glial cells other than oligodendrocytes in vivo. C 2005 V

Wiley-Liss, Inc.

INTRODUCTION Although recent work has partly elucidated the development of oligodendrocytes in the neural tube (Liu et al., 2002; Rowitch, 2004), much remains to be discovered concerning other glial lineages in the rest of the CNS. Our study is aimed at understanding the origin of glial cells in the neurohypophysis (NH), a CNS structure located at the base of the brain (Fig. 1A–D). It is a model of special interest due to its relatively simple cellular composition: apart from axons, blood vessels and microglia, the principal cellular element is the pituicyte. C V

2005 Wiley-Liss, Inc.

Pituicytes are frequently regarded as astroglial cells since, in rat, they express vimentin (Marin et al., 1989) and S-100b (Cocchia and Miani, 1980), two markers of the astrocyte lineage. Glial fibrillary acidic protein (GFAP), another astrocyte marker, is also detected in rat and human pituicytes (Salm et al., 1982; Velasco et al., 1982), but not in mouse pituicytes (our unpublished observation). Pituicytes may thus constitute a specialized population of glial cells distinct from astrocytes, as anticipated from classical histological studies (Bucy, 1932). Importantly, no markers of mature oligodendrocytes are detected in the NH (see Fig. 4A). Nevertheless, an earlier study (Wang et al., 1994) suggested the presence of oligodendrocyte precursor cells (OPCs) in the developing NH: neurohypophysial explants from newborn rats generated migrating cells that exhibited properties similar to those described previously for OPCs from the optic nerve (Raff et al., 1983). During development, OPCs arise from multipotential cells in spatially restricted regions of the germinal zones (Spassky et al., 1998, 2001; Vallstedt et al., 2005). Subsequently, they migrate throughout the CNS to reach their final destinations, where most differentiate into oligodendrocytes, while others persist as OPCs in the adult. Early OPCs express characteristic markers: the proteolipid protein transcripts plp/dm20 (Spassky et al., 1998), which encode major myelin components; the platelet-derived growth factor receptor-a (PDGFRa) (Hall et al., 1996); Sox10, a transcription factor of the high-mobility group domain family, expressed by all glial progenitors (Kuhlbrodt et al., 1998; Zhou et al., 2000); Olig1 and Olig2, two bHLH transcription factors, necessary for early commitment of OPCs and oligodendrocyte lineage specification (Lu et al., 2000; Zhou et al., 2000); Nkx2.2, a homeodomain transcription factor involved in OPC differentiation (Qi

Grant sponsor: Association pour la Recherche contre le Cancer (ARC); Grant sponsor: European Community; Grant number: QLG3-CT-2000-00911; Grant sponsor: INSERM Stem Cell Network. *Correspondence to: Pascale Durbec, Laboratoire de Neurogene`se et Morphogene`se dans le Developpement et chez l’Adulte, CNRS UMR 6156, Universite de la Mediterranee, IBDM, Parc Scientifique de Luminy, 13288 Marseille Cedex 9, France. Email: [email protected] Received 2 May 2005; Accepted 12 August 2005 DOI 10.1002/glia.20282 Published online 1 November 2005 in Wiley InterScience (www.interscience.wiley. com).

OPCS GENERATE PITUICYTES IN VIVO

295

Fig. 1. Chronology of appearance of oligodendrocyte progenitor markers in the developing rat NH. A,C: The hypophysis is located under the brain. Boxes: areas viewed in B and D. B,D: Nissl staining on sagittal sections of E20 and P10 rat hypophysis. NH: neurohypophysis; I: intermediate lobe; AH: adenohypophysis. The pituitary stalk joins the NH to the hypothalamus (Hy). E: At E20, pdgfra1 cells are present throughout the NH (surrounded by the dotted line; inset: larger view of positive cells), where they persist at P0 (F), P10 (G), and in the adult (H). olig11 cells are detected in the NH at E20 (I), but not at its distal end, which is even more noticeable at P0 (J). At P10 (K) and in the adult (L), olig11 cells are restricted to the junction with the pituitary stalk. With the sox10 (M–P) and olig2 (Q–T) probes, we obtain staining patterns quite similar to that of olig1, although somewhat fainter for olig2. In the adult, sox101 cells are present throughout the NH (P). plp/dm20 is not detected at E20 (U), but later (V–X) displays a pattern of expression similar to olig1. Cx, cortex; Cb, cerebellum; OB, olfactory bulb; OC, optic chiasma. Scale bars 5 500 lm in B,D; 250 lm in E–X.

et al., 2001); NG2, a chondroitin sulfate proteoglycan (Nishiyama et al., 1996) and a surface antigen recognized exclusively in vitro by the A2B5 antibody (Raff et al., 1983). Culture of perinatal OPCs from structures such as the optic nerve demonstrated their bipotentiality in vitro: they differentiate into type 2 astrocytes in the presence of fetal calf serum (FCS) and cytokines (Fulton et al., 1991), whereas in serum-free medium they generate oligodendrocytes (Raff et al., 1983; Noble and Wolswijk, 1992; Noble et al., 2004). Other reports suggested that OPCs might actually be bipotential in vivo in pathological situations (reviewed in Franklin and Blakemore, 1995; Franklin, 2002). However, while it is clear that the primary fate of OPCs during CNS development is to generate oligodendrocytes, there is no definitive evidence that they give rise to astrocytes or other

glial cells in vivo (Skoff, 1990; Fulton et al., 1992; Franklin and Blakemore, 1995; Franklin, 2002; Noble et al., 2004). In the present work, we further characterize the cells previously identified as putative OPCs in the developing NH (Wang et al., 1994) and study their relationship with pituicytes. By in situ hybridization, immunolabeling, cell culture, and transplantation, we establish that cells with phenotypic and functional properties of OPCs are present in the developing rodent NH before pituicyte differentiation. Moreover, using genetically engineered mice to follow the fate of these precursors, we demonstrate in vivo that pituicytes derive from these OPCs. We conclude that OPCs generate pituicytes during NH development. This is the first demonstration that OPCs actually give rise to a glial cell population distinct from oligodendrocytes in vivo.

296

VIRARD ET AL.

MATERIALS AND METHODS Solutions for cell culture were purchased from Invitrogen (Cergy Pontoise, France) and other chemicals from Sigma (Saint Quentin Fallavier, France), if not otherwise specified.

Animals and Tissue Processing All procedures involving the use of animals were performed in accordance with the European Animal Care Guidelines and Directives. We used Wistar rat and Swiss mouse embryos and pups from embryonic day 16 (E16) to postnatal day 10 (P10), and 6- to 8-week-old males. Dr. D. Rowitch’s laboratory (Dana-Farber Cancer Institute) provided tissues from the progeny of an olig1cre heterozygote mouse (Lu et al., 2002) mated to a ROSA26 strain carrying a floxed-STOP allele of lacZ (Soriano, 1999). Shiverer mice were obtained from the Jackson Laboratory (Bar Harbor, ME). Brains and hypophyses from animals perfused with 4% paraformaldehyde (PF) were postfixed with 4% PF for 2 h, cryoprotected, and included in OCT (Sakura Finetek, Bayer Diagnostic). Cryostat sections (12 lm) were collected on SuperFrost slides (VWR, Fontenay-sous-Bois, France).

Neurohypophysial Explant Cultures NH culture was performed as described by Wang and colleagues (1994). Briefly, NHs from P0–P3 rats were carefully dissected from the meninges, adenohypophysis, and intermediate lobe. Each was then sectioned into 4–6 pieces and explanted onto poly-L-lysine-treated glass coverslips either in Dulbecco’s modified Eagle’s medium (DMEM) with 10% FCS or in serum-free medium (DMEM-F12 complemented with 100 lg/ml human transferrin, 5 lg/ml insulin, 100 lM putrescin, 20 nM progesterone, 30 nM sodium selenite, and 1 mM sodium pyruvate) supplemented or not with FGF-2 (10 ng/ml), PDGF-AA (10 ng/ ml), and NT3 (10 ng/ml) in the presence of 0.4% methylcellulose. For cell migration assays, A2B51 cells were counted around NH explants in three independent experiments with at least five explants per condition. In experiments designed to analyze the role of proliferation, 10 mM b-D-arabinofuranosylcytosine (ARA-C) was added to the culture medium after 17 h. Cell progression out of the explants was then monitored at different time points (17, 23, and 39 h after explantation) using a Leica DM IRB inverted microscope.

Dissociated and Sorted Cell Cultures P0–P3 rat NHs were dissected, incubated in 5 mg/ml trypsin in Hanks’ balanced salt solution (HBSS) and dissociated by trituration. After rinsing in DMEM containing 10% FCS, the suspension was incubated with the A2B5 antibody (ascites diluted 1:100; American Type

Cell Collection [ATCC], Manassas, VA), then with an anti-mouse IgM antibody conjugated to microbeads (Miltenyi Biotec, Paris, France). Magnetic-activated cell sorting (MACS) was performed according to the manufacturer’s recommendations by passing the cells over an MS cell separation column (Miltenyi Biotec) twice. First, to confirm the identity of the acutely dissociated A2B51 cells, their antigenic phenotype was assayed by rapid centrifugation onto glass slides using a Shandon 4 Cytospin (Thermo, Jouan, Saint-Herblain, France) and by fixation and staining with A2B5 or the anti-Olig2 antibody. Second, to study their differentiation potential, the positive cell fraction was seeded at a concentration of 4,000 cells in 12 ll of DMEM complemented with 10% FCS onto poly-L-lysine-treated glass coverslips and left to settle for 1 h. We then filled the dishes with either DMEM complemented with 10% FCS or the defined medium described above, diluting the serum down to a final concentration of 1%. Both explant and dissociated cell cultures were kept at 37°C in 5% CO2 and 95% air, and one-half of the medium was changed every other day.

Transplantation of Mouse NH Pieces into Shiverer Brain We dissected NH pieces (150–200 lm in diameter) from newborn mice expressing the green fluorescent protein (GFP) under the control of the actin promoter (actin-GFP mice) (Hadjantonakis et al., 1998). We excluded the intermediate lobe and pituitary stalk and only used the part of the NH most distal from the stalk. We then grafted 2–3 NH pieces into the brain of 2-month-old myelin-deficient Shiverer mice (n 5 6) (Lachapelle et al., 1983–1984), within the lateral ventricle. We sacrificed host mice at various times after transplantation (25 or 39 days) and processed their brains for cryostat sectioning, as described above. Sagittal sections (14 lm) were labeled with an anti-MBP antibody. Immunolabeling and Nissl and XGal Staining We used the following primary antibodies: A2B5 (mouse IgM; pure supernatant; ATCC), anti-Olig2 (rabbit; 1:20,000; gift from Dr. Rowitch), polyclonal anti-NG2 (rabbit; 1:200–1:1,000; Chemicon, Euromedex, Mundolsheim, France), monoclonal anti-NG2 (mouse IgG; 1:25; Chemicon, Euromedex) and anti-Nkx2.2 (mouse IgG; 1:800; Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA) for OPCs; O4 (mouse IgM; 1:1; ATCC) for oligodendrocytes and type 2 astrocytes; anti-MBP (mouse IgG; 1:50; Euromedex) and GalC (mouse IgG; 1:1; ATCC) for oligodendrocytes; and antivimentin (clone LN-6; mouse IgM; 1:200; Sigma), anti-S100 (rabbit; 1:300; DAKO, Trappes, France), monoclonal anti-GFAP (mouse IgG; 1:400; Sigma), and polyclonal anti-GFAP (rabbit; 1:100; Sigma) for pituicytes or astrocytes. For immunolabeling, the tissue sections were

OPCS GENERATE PITUICYTES IN VIVO

incubated overnight at 4°C in a primary antibody solution, with 0.1% Triton X-100 for internal markers (GFAP, MBP, Nkx2.2, Olig2, S-100, vimentin). The sections were subsequently washed and incubated with the appropriate fluorescently labeled secondary antibodies (1:1,000; Jackson ImmunoResearch Laboratories, West Grove, PA). To label cultures, we first fixed them for 10 min in 4% PF in phosphate-buffered saline (PBS) and incubated them in the primary antibody for 1 h at room temperature (RT); then, after washes, we soaked them in the corresponding secondary antibody for 1 h at RT. When necessary, we stained cell nuclei with Hoechst 33342. Sections, tissue pieces, and cultures were examined using Zeiss Axiophot fluorescence or confocal microscopes. Nissl and XGal stainings on 12-lm cryosections were performed using classical protocols. To carry out immunohistochemistry after either in situ hybridization or XGal staining, we postfixed the tissue sections with 4% PF and followed the immunohistochemical staining protocol of the Vectastain universal Elite ABC kit (Vector Laboratories, Burlingame, CA). To analyze the glial cells of the adult NH, the number of positive cells for a given marker was counted on at least 5 randomly-chosen sections from adult rat NH and compared with the total number of cells identified with the nuclear marker. When XGal staining was followed by immunohistochemistry, we first counted cells stained with XGal and then performed immunohistochemistry, analyzed the exact same fields, and included in our analysis only cells with a clear immunolabeling.

In Situ Hybridization on Sections Nonradioactive in situ hybridization was done as previously described (Tiveron et al., 1996), with minor modifications. Cryosections (12 lm) of adult rat NH were hybridized with 500 ng/ml digoxigenin-labeled antisense riboprobes for olig1 and olig2 (Lu et al., 2000), pdgfra (Mudhar et al., 1993), sox10 (Kuhlbrodt et al., 1998), or plp/dm20 (Peyron et al., 1997). To improve final color development, 5–10% polyvinyl alcohol was added to the development solution.

RESULTS Glial Progenitor Markers Are Present in the Developing Rat NH To determine whether there are OPCs in the NH, as had been previously suggested (Wang et al., 1994), we performed in situ hybridization with different classical glial markers on developing and adult rat NH. At early stages, OPCs can be identified by the expression of markers such as plp/dm20 (Spassky et al., 1998), plateletderived growth factor receptor a (PDGFRa) (Hall et al., 1996), and the transcription factors Sox10 (Kuhlbrodt et al., 1998; Zhou et al., 2000), Olig1, and Olig2 (Lu et al., 2000; Zhou et al., 2000). In rat, the anlage of the NH is formed before E17.5 as a mass of cells derived

297

from the neuroectoderm. Ultrastructural studies had suggested that cell differentiation in the NH starts from E17.5 and is maintained till one month after birth (Galabov and Schiebler, 1978). In our study, we used tissue sections from E16 to adult rats to investigate the chronology in which glial progenitor markers appear during NH development (Fig. 1). The earliest detection of OPC markers in the anlage of the NH (Fig. 1A,B) was at E20; at this stage, we observed expression of pdgfra (Fig. 1E), olig1 (Fig. 1I), sox10 (Fig. 1M), only weak staining for olig2 (Fig. 1Q) and no staining for plp/dm20 (Fig. 1U). At P0, pdgfra was expressed in cells dispersed throughout the NH (Fig. 1F), while olig1-expressing cells were seen in the stalk and the proximal half of the NH (Fig. 1J). sox10 (Fig. 1N), olig2 (Fig. 1R), and plp/dm20 (Fig. 1V) probes had expression patterns similar to olig1. At later stages, namely at P10 (Fig. 1C,D) and in the adult, pdgfra (Fig. 1G,H) remained expressed in cells located in the entire NH, whereas olig1 (Fig. 1K,L), olig2 (Fig. 1S,T), and plp/dm20 (Fig. 1W,X) were only detected at the junction with the pituitary stalk. With the sox10 probe, at P10, we obtained a clear staining in the stalk and proximal part of the NH as well as in a few cells at the distal end of the NH (Fig. 1O) and, in the adult, a weak staining in cells dispersed throughout the NH (Fig. 1P). To characterize further the cells expressing OPC markers, we carried out double stainings at P0. At this stage, 100% of the olig11 cells in the NH also expressed Olig2 and vice versa (Fig. 2A). Furthermore, all the Olig21 cells were also positive for pdgfra (Fig. 2B) and sox10 (Fig. 2C), as well as for Nkx2.2 (Fig. 2D), another OPC marker (Qi et al., 2001). Thus, at P0, cells in the NH simultaneously, express olig1, Olig2, pdgfra, sox10, and Nkx2.2 a combination of markers also found in OPCs (Liu et al., 2002). Strikingly, the appearance of these cells coincided with the presence of migrating OPCs in the nearby optic chiasma (Fig. 2E,F) (Small et al., 1987). Overall these results demonstrate that progenitor cells expressing concomitantly a set of OPC markers, i.e., olig1/Olig2/pdgfra/sox10/Nkx2.2 are present in the developing NH. To establish that these progenitors are actual OPCs, we also characterized these cells functionally in vitro.

NH Precursors Are Migratory and Bipotential Previous studies showed that OPCs from the optic nerve display two major functional properties in vitro: they are migratory and bipotential, giving rise either to oligodendrocytes or to type 2 astrocytes, depending on the culture medium composition (Raff et al., 1983). Therefore, we first assayed the migratory behavior of NH cells. Newborn NHs were dissected out and cultured in defined medium in the presence of FGF-2, PDGF-AA, and NT3, known to favor the migration, proliferation, and survival of optic nerve OPCs (McKinnon et al., 1990; Barres et al., 1994). Migrating cells were observed around the explants within 24 h when FGF-2, PDGF-

298

VIRARD ET AL.

Fig. 2. Phenotypic characterization of rat NH glial progenitor cells at P0. In situ hybridizations combined with an anti-Olig2 staining demonstrate that the Olig21 cells in the NH are also positive for olig1 (A), pdgfra (B), and sox10 (C). D: Olig21 cells co-express Nkx2.2 as well, as seen by a double immunofluorescent staining. Thus, at P0, cells in the NH express a combination of markers characteristic of OPCs. E: Location of the area shown in F. F: Interestingly, olig11 cells are present in the NH at a time when olig11 OPCs enter the nearby optic chiasma (OC). Scale bars 5 20 lm in A–D; 500 lm in F.

AA, and NT3 were present in the medium. Cell migration also occurred in the presence of the antimitotic agent ARA-C (Fig. 3A–C), indicating that this effect was not due to proliferation. It did not result from explant spreading either since the explants maintained their original shape over time (Fig. 3A–C). The migrating cells were identified using Olig2 as well as NG2 and A2B5, classical in vitro markers for OPCs. We quantified the number of A2B51 cells around the explants after 48 h in the absence or presence of the trophic factors: only rare A2B51 cells were found in the control conditions (