Development 109, 691-698 (1990)Printed in Great Britain © The Company of Biologists Limited 1990

691

Coexistence of perinatal and adult forms of a glial progenitor cell during

development of the rat optic nerve

GUUS WOLSWIJK12, PETER N. RIDDLE3 and MARK NOBLE1

^Ludwig Institute for Cancer Research, 91 Riding House Street, London WIP8BT, UK2lnstitute of Neurology, National Hospital, Queen Square, London WCJN 3BG, UK3Imperial Cancer Research Fund, Lincoln's Inn Fields, London WC2A 3PX, UK

Summary

We have studied the developmental appearance of theO-IA^"" progenitor cell, a specific type of oligodendro-cyte-type-2 astrocyte (O-2A) progenitor cell that we haveidentified previously in cultures prepared from the opticnerves of adult rats. O-2A'"'''" progenitors differ fromtheir counterparts in perinatal animals (O-2Aperina"''progenitor cells) in antigenic phenotype, morphology,cell cycle time, rate of migration, time course of differen-tiation into oligodendrocytes or type-2 astrocytes andsensitivity to the lytic effects of complement in vitro. Inthe present study, we have found that O-2A'"'"" progeni-tor-like cells first appear in the developing optic nerveapproximately 7 days after birth and that by 1 monthafter birth these cells appear to be the dominant progeni-tor population in the nerve. However, the perinatal-to-adult transition in progenitor populations is a gradual

one and O-2Afldu" and O-IA^1"""' progenitors coexist inthe optic nerve for 3 weeks or more. In addition, cellsderived from optic nerves of P21 rats express character-istic features of O-2fdu" and O-2\perina"'! progenitors forextended periods of growth in the same tissue culturedish. Our results thus indicate that the properties thatdistinguish these two types of O-2A progenitors fromeach other are expressed in apparently identical environ-ments. Thus, these cells must either respond to differentsignals present In the environment, or must respondwith markedly different behaviours to the binding ofidentical signalling molecules.

Key words: adult, astrocyte, cell cycle, central nervoussystem, development, differentiation, glia, oligodendrocyte,optic nerve, progenitor, transition.

Introduction

The exponential increases in cell number seen duringembryonic and early postnatal development would beclearly inappropriate if continued into adulthood,where extensive generation of new cells is only requiredin certain lineages (e.g. the lineages of the gut and thehaematopoietic system) or in response to injury. There-fore, developmental mechanisms must exist that allowadult tissues to maintain pools of cells that have theability to participate in regeneration but which normallygenerate new cells at a slow rate, so as to compensatefor naturally-occurring cell death. What are thesemechanisms and how is the transition from the embry-onic to the adult strategy of division and differentiationmodulated within a tissue?

In our studies of the oIigodendrocyte-type-2 (O-2A)astrocyte lineage of the rat optic nerve, we haverecently found that perinatal and adult progenitor cellsdiffer in many properties (Wolswijk and Noble, 1989;Wren and Noble, 1989; for summary see Table 1). Atleast some of these differences appear to representphenotypic specializations relevant to the differing rolesof these cells in developing and mature animals. For

example, O-IM1"""""' progenitor cells have an 18 h cellcycle (Noble et al. 1988) and migrate rapidly in vitro(Small et al. 1987), properties appropriate to the appar-ent need for these cells to populate white matter tractsby migration during embryogenesis and to producesufficient numbers of progenitors rapidly enough toallow myelination to be completed over a relativelyshort time span. In contrast, O-2A"d"'' progenitorsmigrate in vitro with an average speed of only 4 (.im h"1

(Wolswijk and Noble, 1989), less than 20% of the21fimh~1 average speed of O-2Aper'"n'a' progenitors(Small et al. 1987). Moreover, O-2Aflrfu" progenitorshave a greatly lengthened cell cycle time of 65 h(Wolswijk and Noble, 1989). The long cell cycle timesand slow migration rates of O-2Anrf"'' progenitors wouldappear to make these cells more suitable thanQ.2Aperi'""al progenitors for participation in the slowreplacement of oligodendrocytes and type-2 astrocytesthat may be required in the densely packed adult centralnervous system (CNS).

In the present studies, we have addressed two ques-tions central to our understanding of O-2Aarf"'' progeni-tors. First, we determined the time of appearance ofthese cells in the optic nerve and found that between 1

692 G. Wolswijk, P. N. Riddle and M. Noble

Table 1. Defining characteristics of 0-2Ap""""al and0-2Aad"" progenitor cells

Characteristic

O4 labelling (in vitro)Vimentin IFsMorphologyCell cycle timeAverage rate of migration

(on PLL)Time-course of

differentiation(50 % differentiated)

Sensitivity to complement

Source of O-2A

Perinatal opticnerves

- /+ '+c

Bipolar11

18±4he

21.4±1.6/anh~If

<*2 days»-h

i

progenitor cells

Adult opticnerves

+b

_ b

Unipolarb

65±18hb

4.3±0.7/*mirlb

3-4 daysb'h

+'

"Virtually all of the O-2A progenitor cells isolated from the opticnerves of newborn rats are O4~. However, many O-2A progenitorsisolated from the optic nerves 7-day-old rats are initially O4+, butbecome O4~ during the first 24 h in culture when grown in Astro-CM (I. Sommer et al. unpublished data).

bWolswijk and Noble (1989).°Kaffetal. (1984).''Temple and Raff (1986); Small et al. (1987).'Noble et al. (1988).'Smaller al. (1987)."Raff et al. (1983).hln our more recent experiments, O-2\p"iru"al and O-2A'"'""

progenitor cells appear to differentiate more quickly intooligodendrocytes and type-2 astrocytes than in our previousexperiments.

'Wren and Noble (1989).

week and 1 month after birth the progenitor compo-sition of the rat optic nerve changes from predomi-nantly perinatal-like to predominantly adult-like. Fur-thermore, we provide evidence that perinatal and adultprogenitor cells can coexist in the same environment,suggesting that the perinatal-to-adult transition doesnot occur abruptly as the result of a gross change in theCNS environment. It appears instead that O-2Aadu"and O-2Ape"'""a' progenitors respond to identical en-vironments by expressing different properties. Thus,these cells must either respond to different signalspresent in the environment, or must respond withmarkedly different behaviours to the binding of ident-ical signalling molecules.

Materials and methods

Preparation of primary optic nerve culturesOptic nerve cells derived from postnatal day 0 (P0), P7, P14,P21, 1-month-old and adult (2^8 months) Sprague-Dawleyrats were isolated and cultured as described previously(Wolswijk and Noble, 1989). Briefly, minced optic nerveswere digested in 333 i.u. ml collagenase in L-15 medium for60-90 min, followed by addition of an equal volume of30000i.u. ml"1 trypsin in Ca2+-, Mg2+-free DMEM (DMEM-CMF) and an incubation for 20 min at 37 °C. Followingcentrifugation, the tissue was resuspended and further incu-bated in 15 000 i.u. ml"1 trypsin and 0.27 mM EDTA inDMEM-CMF. After 20 min at 37°C, an equal volume of a

solution of soybean trypsin inhibitor (5200i.u. ml '), bovinepancreas DNAse I (74i.u.ml~') and bovine serum albumin(BSA, faction V, 3.0mgmr') in DMEM was added for10-20min. After a short centrifugation, the supernatant wasreplaced by DMEM supplemented with 10 % heat-inactivatedfetal calf serum, 2mM glutamine and 25j/gmr' gentamicin(DMEM+10%FCS) and the tissue was triturated through a5 ml blow-out pipette and through 25G and 27G hypodermicneedles. For immunolabelling and autoradiography exper-iments, dissociated optic nerve cells were plated onto glasscoverslips (Chance No. 0, 13 mm diameter) precoated witheither poly-L-lysine (PLL; 20^gml~') or irradiated (20 Gray)monolayers of 20000 purified type-1 astrocytes prepared fromP0-P1 rat cerebral cortices as described before (Noble andMurray, 1984). Optic nerve cells derived from P0 to1-month-old rats were plated at a density of 3000-4000 cellsper coverslip, while the cell suspension derived from one pairof adult optic nerves was plated onto 8-15 coverslips. After3-4 h at 37°C, the coverslips were rinsed in L-15 medium andplaced in Falcon 6-well trays (3-4 coverslips per well).Cultures were maintained in DMEM-BS/0.5 %FCS [DMEMsupplemented with 0.5% fetal calf serum and various addi-tives modified from Bottenstein and Sato (1979), as describedpreviously (Raff et al. 1983; Wolswijk and Noble, 1989)]. Twothirds of the culture medium was replaced every 2-3 days.

Clonal analysis studiesTo analyze the clonal expansion of O-2A progenitor cellcolonies, dissociated optic nerve cells derived from P21animals were plated into Falcon tissue culture flasks (25 cm2

surface area) that had been coated previously with irradiatedmonolayers of type-1 astrocytes (750000 cells/flask). Prior toplating, GalC+ oligodendrocytes were eliminated from thecell suspensions by complement-mediated cytolysis. Dis-sociated P21 optic nerve cells were incubated in L-15 mediumcontaining agarose-absorbed rabbit complement (Buxted Ltd,UK; diluted 1:10) and anti-GalC antibodies (hybridomaculture supernatant, diluted 1:10; Ranscht et al. 1982) for30min at 37CC, followed by addition of a 20-fold volume ofL-15 and a centrifugation at 500 g for 5 min. The pellet wasresuspended in DMEM-BS/0.5 %FCS and 1000 of the re-maining cells were plated per 25 cm2 flask. From day 1onwards, cultures were screened for colonies of O-2A pro-genitor-like cells. When a colony (3=2cells) was found, itslocation was marked and its expansion was then followeddaily using an Olympus inverted microscope. As each flaskcontained fewer than 20 colonies of O-2A progenitor-like cells(i.e. less than 1 colony per 125 mm2), it seems likely that eachcolony was derived from a single cell. The appearance ofmultipolar oligodendrocyte-like cells in the colonies con-firmed that those colonies initially followed contained O-2Aprogenitor cells. Colonies of live O-2A progenitors growingon monolayers of type-1 astrocytes were photographedthrough an inverted microscope using Kodak Tri-X-pan 125ASA films.

Time-lapse microcinematographyDissociated optic nerve cells derived from P21 animals weredepleted of GalC+ oligodendrocytes by complement-me-diated cytolysis and seeded at a density of 5000-10 000 cells inthe centre of PLL-coated 6 cm Nunc Petri dishes of which theedges had been coated previously with 250000 purified type-1astrocytes (to condition the medium) as described before

Coexistence of 0-2Aperinatal and 0-2Aaduh progenitors 693

(Small et al. 1987; Noble et al. 1988; Wolswijk and Noble,1989). Two hours later, cultures were rinsed with L-15medium (to remove debris and myelin), followed by additionof 50% DMEM-BS/0.5%FCS that had been conditioned bytype-1 astrocytes (for 24-48 h) plus 50% fresh DMEM-BS/0.5%FCS. Cultures were placed in the time-lapse microcine-matography set-up [Olympus inverted microscopes adaptedfor time-lapse microcinematography (Riddle, 1979)] 1 dayafter plating. Fields were chosen that contained at least 3O-2A progenitor-like cells. Photographs were recorded every300 s and 16 mm Kodak Infocapture AHU microfilm 1454films were used. The distance an O-2A progenitor-like cellmigrated between divisions was determined using a HewlettPackard 9874A digitizer.

Antibodies and immunofluorescence stainingAll the antibodies and the experimental procedures that wereused in our experiments have been described previously: the]gM monoclonal antibody A2B5 (hybridoma supernatant,diluted 1:1; Eisenbarth et al. 1979), the IgM monoclonalantibody O4 (concentrated hybridoma supernatant, 1:100;Sommer and Schachner, 1981), the lgG3 monoclonal anti-GalC antibody (hybridoma supernatant, 1:10; Ranscht et al.1982), the IgGi monoclonal anti-vimentin antibody(Boehringer Mannheim GmbH, West-Germany, used at aconcentration of 4 /.igml"') and rabbit anti-GFAP antiserum(diluted 1:1000; Dako Ltd., UK). Binding of the first-layerantibodies was visualized using fluorescein (Fl)- or rhodamine(Rd)-coupled goat anti-mouse IgM (for A2B5 and O4 anti-bodies), goat anti-mouse IgGpFl (for anti-vimentin anti-bodies), goat anti-mouse IgG3-Fl (for anti-GalC antibodies)and sheep anti-rabbit Ig-Fl (for anti-GFAP antibodies). Allsecond-layer antibodies were purchased from Southern Bio-technology Associates, Inc., USA and diluted 1:100 prior touse. Antibodies were diluted in Hanks' balanced salt solutionbuffered with 0.02 M Hepes and supplemented with 5 % heat-inactived donor calf serum and 0.05 % (w/v) sodium azide(Sigma) (HBBS+5%DCS). Cultures were incubated in theantibody solutions for 20-30 min at room temperature andthen rinsed several times in HBBS+5%DCS before the nextlayer was applied. Surface immunolabellings (with the mono-clonal antibodies A2B5, 04 and anti-GalC) were carried outon live cells, while intracellular antigens (GFAP and vimentinIFs) were visualized after fixation in methanol at —20°C for10min. In addition to the standard double immunolabellingprocedures, one triple immunolabelling was performed. Toidentify optic nerve cells that were O4+GalC~ and lackedvimentin intermediate filaments, cultures were incubated withanti-GalC antibody, anti-IgG3-Fl, monoclonal 04 antibody,and anti-lgM-Rd, fixed and then labelled with monoclonalanti-vimentin antibodies and anti-IgG|-Fl. In these exper-iments, cells that were surface fluorescein+ (i.e. were GalC+)were easily distinguishable from cells that were internallylabelled by the fluorescein conjugate (i.e. were vimentin"1").After the immunolabelling, coverslips were washed withHBBS+5%DCS and distilled water, mounted in a drop ofglycerol containing 22 mM 1,4-diazobicyclo [2,2,2] octane toprevent fading (Johnson et al. 1982; Davidson and Goodwin,1983), sealed with transparent nail varnish and examined on aZeiss Universal microscope. All immunolabelled cells incultures of adult optic nerve were counted; in culturesprepared from the optic nerves of P0 to 1-month-old animals,at least 200 immunolabelled cells were scored per coverslip ineach experiment. Immunolabelled cells were photographedusing Ilford Xpl 400 films.

Results

Cells with the antigenic and morphologicalcharacteristics of O-2Aadu" progenitor cells appear inthe optic nerve approximately 1 week after birth andbecome the dominant O-2A progenitor population bythe age of 1 monthAs judged by antigenic criteria, O-2Aarf"'' progenitor-like cells were not detectable in cultures prepared fromoptic nerves of newborn rats, were present at lowfrequency (<5 %) in the optic nerves of postnatal day 7(P7) rats, and represented almost 70% of all O-2Aprogenitors in cultures derived from the optic nerves of1 month old rats. O-2Ap*ri"alal progenitors containvimentin intermediate filaments and are A2B5+O4"when induced to divide by type-1 astrocytes (I. Sommeret al. unpublished data). In contrast, O-2A" progeni-tors are both A2B5+ and O4+, but lack intermediatefilaments after 1 day of in vitro growth (Wolswijk andNoble, 1989).

In cultures prepared from the optic nerves of P0 rats,less than 1 % of the O-2A progenitors (as defined by anA2B5+GalC~ antigenic phenotype) were labelled withthe 04 antibody after 1 day of growth in these cultureconditions, as expected for O-2A/7er""2'a/ progenitors (I.Sommer et al. unpublished data). This percentage roseto 78±1 % in P14 cultures, and by 1 month after birth>95 % of such cells were O4+ when grown in identicalconditions (Fig. 1). Furthermore, the proportion of theO4+ O-2A progenitors that, like O-2Aarfu)' progenitors,lacked vimentin IFs after 1 day in culture increasedfrom sS5% in P7 optic nerve cultures to 68±6% incultures prepared from the optic nerves of 1-month-oldrats (Fig. 1). In addition, the amount of vimentinfilaments in O4+ O-2A progenitor cells that werevimentin"1" diminished in correspondence with the ageof the rats, and many of such cells only had vimentinfilaments in the tips of their processes (data not shown).These two sets of immunolabelling experiments allowedus to determine the proportion of the total O-2Aprogenitor population that had the O4+ vimentin"antigenic phenotype of O-2And"/' progenitors (Fig. 1C).In cultures prepared from P7 rat optic nerves,1.8±0.4% of the O-2A progenitor population wereO4+ and devoid of vimentin IFs, while in optic nervecultures of 1-month-old animals 67±5% of the totalO-2A progenitor population had the O4+vimentin~antigenic phenotype of O-2A"rfu/' progenitors.

The morphologies of cultured O-2A progenitorsderived from P0-1 month-old rats also differed inaccord with the view that O-2Aadu/' progenitors becameincreasingly frequent during the first month after birth.We have previously shown that O-2A progenitors incultures of adult optic nerve are unipolar cells(Wolswijk and Noble, 1989), while most O-2A progeni-tors in cultures prepared from the optic nerve from E17to P14 are bipolar in their morphology (Temple andRaff, 1986; Small et al. 1987). In optic nerve culturesderived from 1 month old rats, 32±2% of O-2Aprogenitors (as defined by an A2B5+GalC~ GFAP"antigenic phenotype) could be classified clearly asunipolar cells after 3 days in Astro-CM, 14±5 % had

Coexistence of 0-2 AP*™atal and O-2Aadult progenitors 695

Fig. 2. An O4+GalC vimentin 0-2A progenitor cell in a P14 optic nerve culture with a morphology similar to that of anO-2Aarf"'' progenitor cell. Optic nerve cells derived from P14 animals were plated onto PLL-coated coverslips and grownovernight in type-1 astrocyte conditioned DMEM-BS/0.5%FCS. Cells were then triple immunolabelled with anti-GalCantibodies, 04 antibodies, followed by fixation and labelling with anti-vimentin antibodies as described in Materials andmethods. This figure illustrates that cultures of perinatal optic nerve contain O-2A progenitors with the morphology andantigenic phenotype of 0-2A'"'"'' progenitor cells, i.e. unipolar O4+GalC~vimentin~ cells. One such cell is indicated with anarrow. A GalC+O4+ oligodendrocyte is indicated with an arrow head. (A) Phase-contrast optics; (B) rhodamine optics;(C) fluorescein optics. Bar in A, 50 ̂ m.

number of these cells was too low to allow meaningfulquantification. Interestingly, in the light of the long cellcycle times of O-2A° progenitors (Wolswijk andNoble, 1989) and rapid cell cycle times of O-2Ape"na"j/

progenitors (Noble et al. 1988), the adult-like unipolarO-2A progenitors in cultures of optic nerve of 1 monthold rats were most frequently found as single cells orgroups of 2 cells, while the perinatal-like O-2A progeni-tor cells were found predominantly in groups of =S8 cells(data not shown).

Cells with the cell cycle time, rate of migration andmorphology of O-2Aper"""al and O-2Aadu" progenitorcells co-exist in P21 optic nerve culturesHaving found that cells with the characteristic antigenicphenotype and morphological features of bothO-2Aa^" and O-2Apcn'""al progenitors could be ident-ified in cultures derived from nerves of young rats, wenext determined whether these cultures also containedcells that expressed the markedly different cell cycletimes and migratory behaviours of these two O-2Aprogenitor populations. We conducted our experimentson cell division and migration on cells derived from P21rats. This age was chosen as midway between the agesof P14 (when there were few cells with the antigens andmorphology of O-2A'"'"" progenitors) and 1 monthafter birth (when there were relatively few cells with theantigens and morphology of O-2A''er"""al progenitors).

When we plated cells derived from optic nerves ofP21 rats onto type-1 astrocyte monolayers at a densityof 1 cell per 25 mm2, we found colonies that expandedin a manner characteristic of O-2Aorfu" progenitors, andother colonies that expanded in a manner characteristicof O-2Aprn"°'a/ progenitors. We initially followed the

expansion of 35 individual, well-isolated clones of O-2Aprogenitor-like cells on a daily basis. All colonies werefollowed identically, with note made of cell number andcellular morphology. In our final analysis, colonies thatalready contained oligodendrocyte-like cells when firstnoted were not included. This culling left us with 18colonies suitable for analysis.

As shown in Fig. 3, 9/18 of the colonies analyzedexpanded with an average doubling time of 68±11 h, arate similar to that of colonies of O-2Aarf"'' progenitorcells isolated from the optic nerves of adult rats(Wolswijk and Noble, 1989). Such slowly expandingcolonies consisted initially of unipolar, O-2Aod"'r pro-genitor-like cells (Fig. 4). In contrast, 9/18 of thesecolonies expanded until the first appearance of multi-polar oligodendrocyte-like cells (after 6-9 days in vitroin these particular colonies), with an average doublingtime of 29±4h, which is comparable to the 24h doub-ling time of O-2Apen'"a'fl/ progenitor colonies (Nobleand Muiray, 1984; Temple and Raff, 1986). The O-2Aprogenitor-like cells present in the rapidly expandingcolonies expressed the characteristic bipolar mor-phology of O-2A progenitor cells derived from the opticnerves of perinatal rats (Fig. 4). The differences inpopulation doubling times between the rapidly expand-ing, perinatal-type colonies and the slowly expanding,adult-type colonies is reflected in marked differences incell number. For example, expanding perinatal-typecolonies contained 28±15 cells (n=7) after 7 days invitro, while expanding adult-type colonies contained onaverage only 8±4 cells (n=9). When multipolar oligo-dendrocyte-like cells appeared in either type of colony,the doubling time of the colony declined (Fig. 3).However, this decline in doubling time occurred much

696 G. Wolswijk, P. N. Riddle and M. Noble

zoouz

u

cHIma

o

8

3u

UJoX

Oo

at

UJo

a.um

s

100-3

10 -

S 1 0 1 6

NUMBER OF DAYS IN CULTURE

100 -i

10 -

B

5 10 15

NUMBER OF DAYS IN CULTURE

100-j

10:

P21 PERINATAL-TYPEP21 ADULT-TYPEADULT

S 10 1 SNUMBER OF DAYS IN CULTURE

more rapidly in the perinatal-type colonies than in theadult-type colonies (Fig. 3). After 15 days in vitro,virtually all of the cells in the perinatal-type colonieshad an oligodendrocyte-like morphology, while theadult-type colonies contained variable proportions ofoligodendrocyte-like cells and O-2Aadu/' progenitor-likecells (data not shown).

Although the above data indicate that optic nerves ofP21 rats contain cells that produce colonies able toexpand with rates characteristic of perinatal or adultprogenitors, they do not give information about the cellcycle times of individual cells. To obtain such infor-mation, and also to examine migratory characteristicsof these cells, we followed O-2A progenitor cells in P21optic nerve cultures using time-lapse microcinema-tography. In these experiments, optic nerve cells de-rived from P21 rats were grown on poly-L-lysine-coatedtissue culture plastic in the presence of Astro-CM for3-8 days.

In our analysis of 73 individual O-2A progenitor-likecells derived from P21 animals, we found that O-2Aprogenitor-like cells that had a division time of <30h

Fig. 3. Cultures of P21 rat optic nerve contain colonies ofO-2A progenitor cells that expand at rates similar to that ofO-2A'"r"""a' progenitor cells and colonies that expand atrates similar to that of O-2A"d"" progenitor cells.Dissociated optic nerve cells derived from P21 rats anddepleted of GalC+ oligodendrocytes were plated at lowdensity (1 cell per 2.5 mm2) into flasks coated withirradiated monolayers of type-1 astrocytes. Cultures werescreened for colonies of O-2A progenitor-like cells. Whenfound, their location was marked and their expansion wasthen followed daily. The figure depicts the expansion ofcolonies which, on the basis of their doubling time, cellnumber and on the basis of the morphology of the O-2Aprogenitor cells in these colonies, were classified aso_2AP«i»atai progenitor-type colonies (A) or O-2Aad""progenitor-type colonies (B).

(A) 9 of the 18 colonies (colonies a-i) analyzedexpanded, up until the appearance of multipolaroligodendrocyte-like cells, with an average doubling time of29±4h (determined using regression analysis) andcontained cells that looked morphologically likeQ_2fij"nnatai p r o g e n j t o r c ens (see pig. 4). When themajority of the cells in a particular colony had acquired amultipolar, oligodendrocyte-like morphology, the expansionof that colony was sometimes not followed further. Notethat the appearance of multipolar oligodendrocyte-like cellsin these perinatal-type colonies (after 6-9 days in vitro)resulted in a rapid decline of the doubling time, consistentwith the observation that clonally related O-2A'*"'"0"1'progenitor cells generally differentiate into oligodendrocyteswithin a short time of each other (Temple and Raff, 1986;Dubois-Dalcq, 1987; Wren et al. unpublished data).

(B) The remaining 9 colonies (colonies A-l) expanded ata much slower rate. Initially, these colonies increased insize with an average doubling-time of 68±llh, i.e. at a ratesimilar to that of colonies of O-2A'"'"'' progenitor cells.Such colonies consisted initially entirely of unipolarO-2Aod"'( progenitor-like cells (see Fig. 4). Note that whenmultipolar, oligodendrocyte-like cells appeared in a colony,the doubling-time of that colony declined, but moregradually than was the case for colonies of O-2AA"rr'"a"1'progenitor cells.

(C) Summary of data presented in A and B. At eachtime-point, the average number of cells in the perinatal-and adult-type colonies was determined and the resultinggraphs were compared to a similar graph constructed usingdata from colonies of O-2A"rf"'' progenitors growing incultures of adult optic nerve (from Wolswijk and Noble,1989). The figure shows that the average cell numbers ateach time point and the rate of expansion of the adult-typecolonies in cultures of P21 optic nerve are identical tocolonies of O-2Aad"" progenitors in cultures of adult opticnerve and that these results are very different from theaverage cell numbers and rate of expansion of the perinatal-type colonies in cultures of P21 optic nerve.

migrated at a rate of 18.6±7.7/imh x («=45), whileO-2A progenitor-like cells that had a division time of>40h migrated at a rate of 7.2±2.5/umh~I («=14)(Fig. 5). In addition, the rapidly dividing and migratingO-2A progenitor-like cells were generally bipolar,while the slowly dividing and migrating O-2A progeni-tor-like cells were generally unipolar. Of the cells withintermediate cell cycle times (i.e. between 30 and 40h),expressed intermediate rates of migration (average

Coexistence of Q and 0-2Aadult progenitors 697

Fig. 4. O-2A progenitor cellsin rapidly expanding coloniesare predominantly bipolar,while such cells in slowlyexpanding colonies are mostlyunipolar. Colonies of O-2Aprogenitor cells growing onmonolayers of type-1 astrocyteswere photographed through aninverted microscope after 6days (A) or 10 days (B) invitro. Note that the cells in therapidly expanding colonies (A)are mostly bipolar(arrowheads), whereas those inthe slowly expanding colonies(B) are predominantly unipolar(arrows). Bar in A, 50 fim.

rate: 10.9±4.6jUmh l, n=14), six of these cells wereundergoing their final division prior to oligodendrocyticdifferentiation.

Discussion

Our previous studies have revealed that oligodendro-cyte-type-2 astrocyte (O-2A) progenitor cells isolatedfrom the optic nerves of adult rats differ from their

30-1

O<<a.a

< 30 > 40

DIVISION TIME IN HOURS

Fig. 5. O-2A progenitors with a short cell cycle time andbipolar morphology in cultures of P21 optic nerve migraterapidly, while unipolar slowly dividing O-2A progenitorsare much less motile. Cells growing in cultures of P21 opticnerve were followed using time-lapse microcinematography.O-2A progenitor-like cells with a division time of <30h andthe bipolar morphology of O-2N"nnatal progenitorsmigrated with average speeds of 18.6±7.7^mh~' (n=45).In contrast, O-2A progenitor cells with the slow divisiontime (>40h) and unipolar morphology of O-2Aad""progenitors migrated much more slowly, with averagespeeds of 7.2+2.5/imh"1 (n= 14). O-2A progenitor-likecells with a cell cycle time of 30-40 h had intermediatemigration rates (data not shown).

perinatal counterparts in several respects. We founddifferences in antigenic phenotype, morphology, cellcycle time, rate of migration, time course of differen-tiation into oligodendrocytes or type-2 astrocytes andsensitivity to the lytic effects of complement in vitro(Wolswijk and Noble, 1989; Wren and Noble, 1989; seeTable 1 for summary). The ability to identify O-2Aarf""progenitor cells in vitro has now enabled us to investi-gate the timing of appearance of O-2Andu" progenitorcells during development of the rat optic nerve. Wehave found that cells with the antigenic and morpho-logical characteristics of O-2Aarf"/; progenitor cells firstappear in the optic nerve approximately 7 days afterbirth. These cells coexist with o^A'"""0"1' progenitorsfor several weeks, during which time the proportion ofdifferent progenitor types isolated from the nervegradually shifts from predominantly perinatal to pre-dominantly adult-like. Moreover, in cultures derivedfrom P21 rats, we could identify cells that expressed thelong cell-cycle times, slow migratory rates and unipolarmorphology of O-2Aadu/' progenitors in the same cul-ture dish as cells that expressed the rapid cell cycletimes, rapid migration and bipolar morphology ofO-2Aperini"a' progenitors. Our observations thatO-2AP"ma<a' and O-2Aodu" progenitor cells coexist inthe optic nerve and in tissue culture for prolongedperiods suggest that the properties that define these twocell types in vitro are not directly modulated by differ-ences between the environments found in perinatal andadult optic nerve, but instead represent intrinsicproperties of these two progenitor populations.

There were several criteria that we used collectivelyto identify O-2Aarfu" and O-2Aper"""a/ progenitors incultures prepared from the optic nerves of young rats.Antigenically, cells that were O4+vimentin~GalC~after 1 day of growth in Astro-CM were considered tobe adult-like, while O-2Ap<rr"ia"j/ progenitors wereO4~vimentin+GalCT. In cultures in which most pro-genitors were perinatal-like (by antigenic criteria), theprogenitors were also predominantly bipolar. In con-trast, in cultures derived from 1 month old rats, many

698 G. Wolswijk, P. N. Riddle and M. Noble

progenitors expressed the antigenic phenotype ofO-2Aarfu/' progenitors and many progenitors also ex-pressed the characteristic unipolar morphology ofO-2Aarfu/' progenitors. In addition, in cultures derivedfrom P21 rats, colony expansion rate and morphologycorrelated with each other, such that rapidly expandingcolonies consisted of bipolar perinatal-like cells, whileslowly expanding colonies consisted of unipolar adult-like cells. The division times of individual cells in similarcultures was also correlated with migratory rates invitro, and cells that divided every 30h or less had anaverage migration rate of 18.6 ^mh"1, while cells with acell cycle time of 40 h or more had an average migrationrate of 7.2 /imh"1.

It is important to note that not all cells could beunambiguously identified as O-2Aper"""fl/ or O-2Aadu"progenitor cells. For example, many of the O4+ cells inthe O-2A progenitor population derived from the opticnerves of P14 and 1 month old rats had a multipolarmorphology and were vimentin+. It seems likely thatthese O4+vimentin+ multipolar cells were visualized atstages of development intermediate betweenQ_2AP'rina,ai p r o g e n j t o r c ens and oligodendrocytes (I.Sommer et al. unpublished data). Such intermediatestage cells might also include some of the cells withdivision times between 30-40 h and average migratoryrates of 10.9±4.6^mh~\

Evidence for phenotypic differences between adultand perinatal precursors in other cellular lineages hasalso been reported by other investigators, indicatingthat developmental transitions in precursor populationsalso occur in other tissues. For example, haematopoi-etic stem cells derived from adult animals predomi-nantly generate erythrocytes that synthesize the adultform of haemoglobin, while erythrocytes derived fromfetal haematopoietic stem cells synthesize fetal haemo-globin (reviewed in Wood et al. 1988). This fetal-to-adult switch in haemoglobin type is apparently regu-lated by an internal clock that resides in the haemato-poietic stem cell population and which is not regulatedby the maturation state of the whole animal. Along withthe observation that erythrocytes that separately pro-duce either fetal or adult haemoglobin coexist in vivofor extended periods, fetal haematopoietic stem cellstransplanted into adult animals generate fetal haemo-globin-synthesizing erythrocytes for a period of timeclosely related to gestational age. These results suggestthat the differences between adult and fetal haemoto-poietic stem cells may be inherent to the cells them-selves and not induced by the environment of the adulttissue (Bunch et al. 1981; Wood et al. 1985, 1988). Suchan inherent encoding of phenotype was also apparent inthe O-2A progenitor populations, in that O-2Aarfu" andQ_2Aperhwmi p r O g e n j t o r c ens w e r e abie t o Co-exist and

express their characteristic phenotypes in the sameenvironment. It remains to be determined whetherthese results indicate that fetal and adult forms ofprogenitors respond to different environmental factorsor interpret identical signals in different ways.

Our thanks are due to Carol Gomm and Liz Parsons whohelped us with the time-lapse microcinematography studiesand to Damian Wren, Martin Raff, Jack Price and Ian

McDonald with whom we had stimulating discussions andwho had helpful comments on the manuscript. This work wassupported by grants from the Multiple Sclerosis Society andMedical Research Council of Great Britain (G. W. and M.N.).

References

BOTTENSTEIN, J. E. AND SATO, G. H. (1979). Growth of a ratneuroblastoma cell line in serum-free supplemented medium.Proc. natn. Acad. Sa. U.S.A. 76, 514-517.

BUNCH, C , WOOD, W. G., WEATHERALL, D. J., ROBINSON, J. S.AND CORP, M. J. (1981). Haemoglobin synthesis by fetalerythroid cells in an adult environment. British J. Haematology49, 325-336.

DAVIDSON, R. S. AND GOODWIN, D. (1983). An effective way ofretarding the fading of fluorescence of labelled material duringmicroscopical examination. Proc. Royal Microsc. Soc. 18, 151.

DUBOIS-DALCQ, M. (1987). Characterization of a slowlyproliferative cell along the oligodendrocyte differentiationpathway. EMBO J. 6, 2587-2595.

ElSENBARTH, G. S., WALSH, F. S. AND NlRENBERC, M. (1979).Monoclonal antibody to a plasma membrane antigen of neurons.Proc. natn. Acad. Sci. U.S.A. 76, 4913-4917.

JOHNSON, G. D., DAVIDSON, R. S., MCNAMEE, K. C , RUSSELL, G.,GOODWIN, D. AND HOLBOROW, E. J. (1982). Fading of immuno-fluorescence during microscopy: A study of the phenomenon andits remedy. J. Immunol. Methods 55, 231-242.

NOBLE, M. AND MURRAY, K. (1984). Purified astrocytes promotethe in vitro division of a bipotential glial progenitor cell. TheEMBO J. 3, 2243-2247.

NOBLE, M., MURRAY, K., STROOBANT, P., WATERFIELD, M. D. ANDRIDDLE, P. (1988). Platelet-derived growth factor promotesdivision and motility and inhibits premature differentiation of theoligodendrocyte/type-2 astrocyte progenitor cell. Nature 333,560-562.

RAFF, M. C , MILLER, R. H. AND NOBLE, M. (1983). A glialprogenitor cell that develops in vitro into an astrocyte or anoligodendrocyte depending on the culture medium. Nature 303,390-3%.

RAFF, M. C , WILLIAMS, B. P. AND MILLER, R. H. (1984). The invitro differentiation of a bipotential glial progenitor cell. TheEMBO J. 3, 1857-1864.

RANSCHT, B., CLAPSHAW, P. A., PRICE, J., NOBLE, M. AND SEIFERT,W. (1982). Development of oligodendrocytes and Schwann cellsstudied with a monoclonal antibody against galactocerebroside.Proc. natn. Acad. Sci. U.S.A. 79, 2709-2713.

RIDDLE, P. N. (1979). Time-lapse cinemicroscopy. BiologicalTechniques Series (ed. J. E. Treherne and P. H. Rubery)Academic Press, London.

SMALL, R. K., RIDDLE, P. AND NOBLE, M. (1987). Evidence formigration of oligodendrocyte-type 2 astrocyte progenitor cellsinto the developing rat optic nerve. Nature 328, 155-157.

SOMMER, I. AND SCHACHNER, M. (1981). Monoclonal antibodies (Olto O4) to oligodendrocyte cell surfaces: An immunocytologicalstudy in the central nervous system. Devi Biol. 83, 311-327.

TEMPLE, S. AND RAFF, M. C. (1986). Clonal analysis ofoligodendrocyte development in culture: Evidence for adevelopmental clock that counts cell divisions. Cell 44, 773-779.

WOLSWIJK, G. AND NOBLE, M. (1989). Identification of an adult-specific glial progenitor cell. Development 105, 387-400.

WOOD, W. G., BUNCH, C , KELLY, S., GUNN, Y. AND BRECKON, G.(1985). Control of haemoglobin switching by a developmentalclock? Nature 313, 320-323.

WOOD, W. G., HOWES, S. AND BUNCH, C. (1988). Developmentalclocks and hemoglobin switching. In Hemoglobin Switching,volume V (ed. G. Stamatoyannopoulos), (in press).

WREN, D. R. AND NOBLE, M. (1989). Oligodendrocytes andoligodendrocyte/type-2 astrocyte progenitor cells of adult ratsare specifically susceptible to the lytic effects of complement inthe absence of antibody. Proc. natn. Acad. Sci. U.S.A. 86,9025-9029.

(Accepted 30 March 1990)