Cell, Vol. 96, 737–749, March 5, 1999, Copyright 1999 by Cell PressProspective Identification, Isolationby Flow Cytometry, and In Vivo Self-Renewalof Multipotent Mammalian Neural Crest Stem Cellsmarkingexperimentsinvivo(Bronner-FraserandFraser,1988; Frank and Sanes, 1991), as well as clonogenicassays in vitro (Sieber-Blum and Cohen, 1980; Baroffioet al., 1988), have shown that many individual neuralcrest cells are multipotent. Previously, we described theSean J. Morrison,†Patricia M. White,†Christiane Zock,†and David J. Anderson*†‡*Howard Hughes Medical Institute†Division of Biology 216-76California Institute of TechnologyPasadena, California 91125 isolation of a multipotent progenitor from the mamma-lian neural crest (Stemple and Anderson, 1992). Thesecells have the capacity to self-renew and to generateneurons, glia, and smooth muscle (Shah et al., 1996)Summaryand have therefore been termed neural crest stem cells(NCSCs). As these cells were isolated from neural tubeMultipotent and self-renewing neural stem cells haveexplant cultures, however, there is no evidence thatbeen isolated in culture, but equivalent cells have notmultipotent NCSCs actually self-renew in vivo. To theyet been prospectively identified in neural tissue. Us-contrary, the proliferation of neural crest cells has beening cell surface markers and flow cytometry, we haveassociated with progressive restrictions in develop-isolated neural crest stem cells (NCSCs) from mamma-mental potential (reviewed by Anderson, 1993).lian fetal peripheral nerve. These cells are phenotypi-We sought to prospectively identify and isolate post-cally and functionally indistinguishable from NCSCsmigratory neural crest cells in order to determinepreviously isolated by culturing embryonic neural tubewhether they are multipotent and self-renewing in vivoexplants. Moreover, in vivo BrdU labeling indicatesas well as in vitro. To do this, we searched for such cellsthat these stem cells self-renew in vivo. NCSCs freshlyin fetal peripheral nerve, which is thought to containisolated from nerve tissue can be directly transplantedonly Schwann (glial) cell precursors (Jessen et al., 1994).in vivo, where they generate both neurons and glia.Using antibodies against the cell surface antigens p75,These data indicate that neural stem cells persist inthe low-affinity neurotrophin receptor, and P0, a periph-peripheralnerveintolategestationbyundergoingself-eral myelin protein (Lee et al., 1997), we have fraction-renewal. Such persistence may explain the origins ofated the E14.5 sciatic nerve by flow cytometry into fivesome PNS tumors in humans.distinct subpopulations. Surprisingly, the p751P02sub-fraction was highly enriched in cells that were function-Introductionally indistinguishable from NCSCs in vitro. Freshly iso-lated p751P02cells also generated neurons and gliaStem cells are self-renewing multipotent progenitorsafter direct transplantation into chick embryos. Finally,with the broadest developmental potential in a giveninvivocellcycleanalysis and BrdU incorporationexperi-tissue at a given time (Morrison et al., 1997a). Neuralments indicated that the p751P02cells undergo self-stem cells have aroused a great deal of interest (re-renewing divisions in the sciatic nerve. Taken together,viewedbyCameronandMcKay,1998;Gage,1998;Tem-these data indicate that multipotent neural crest cellsple and Alvarez-Buylla, 1999) not only because of theirself-renew in vivo and persist into late gestation at leastdevelopmental importance but also for their therapeutica week after the onset of neural crest migration. Thepotential (Gage et al., 1995). A major limitation in thepersistence of NCSCs is of potential therapeutic impor-study of neural stem cells has been the inability to iden-tance and may explain the origin of some PNS tumorstify them prospectively in vivo. This is because therein humans.have been no markers to isolate the stem cells or todistinguish them from restricted progenitors in vivo.Thus, multipotent, self-renewing neural stem cells haveResultsall been isolated after a period of growth in culture thatcould change their properties (Stemple and Anderson,The Fetal Sciatic Nerve Contains Multipotent1992; Kilpatrick and Bartlett, 1993; Davis and Temple,and Committed Neural Progenitors1994; Gritti et al., 1996; Johe et al., 1996; Kalyani et al.,To examine their cellular composition, E14.5–E17.5 rat1997; Palmer et al., 1997). It is therefore not yet clearsciatic nerves were dissociated and cultured at clonalwhether such cells derive from cells with similar proper-density (see Experimental Procedures). After 14 days,ties in vivo.the cultures were fixed and analyzed with immunocyto-We have used the neural crest as a model system tochemical markers. We identified three cell types in suchstudy the biology of mammalian neural stem cells. Neu-cultures: neurons, Schwann cells, and smooth muscle–ral crest cells delaminate from the dorsal neural tubelike myofibroblasts. Neurons were typically identifiedand migrate extensively before aggregating to form theby expression of peripherin, a marker of mature PNSganglia and neuroendocrine tissues of the PNS, as wellneurons (Parysek and Goldman, 1988) (Figure 1B), butas mesectodermal tissues such as smooth muscle andalso expressed neuron-specific tubulin (TuJ1) and neu-bone (Le Douarin, 1980, 1982). In avian embryos, clonalrofilament 160 (data not shown). Schwann (glial) cellswere typically identified by expression of glial fibrillaryacidic protein (GFAP) (Figure 1D) but also expressed‡To whom correspondence should be addressed (e-mail c/[email protected]).p75 and cytoplasmic S100b. Although S100b has beenCell738Figure 1. A Multipotent Colony after Culturein Standard MediumDissociated E14.5 rat sciatic nerve cells wereplated at clonal density and cultured understandard conditions for 14 days and thenfixed and immunohistochemically stained.The same field from a typical multipotentclone isshown in eachpanel,including phasecontrast (A), bright field showing peripherinstaining (B), DAPI fluorescence showing nu-clei (C), glial fibrillary acidic protein (GFAP)(D), and smooth muscle actin (SMA) (E). Peri-pherin staining indicates neurons, GFAPstaining indicates Schwann cells (glia), andSMA staining indicates myofibroblasts.used as a Schwann cell marker in previous studies (Jes- A substantial number of colonies contained neurons,Schwann cells, and myofibroblasts (N1S1M; Figure 1).sen and Mirsky, 1992),