Separate proiiferation kinetics of FGF-responsive and EGF- responsive neural stem ceiis within the embryonic forebrain

germinal zone

BY

David J. Martens

Department of Anatomy and Cell Biology Faculty of Medicine

A thesis submitted in conformity witb the requirements for the Degree of Master of Science, Graduate Department of Anatomy and Cell Biology, University of

Toronto

O by David J. Martens 1999

National Library 1*1 of Canada Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliographic Services seiuices bibliographiques

395 Wellington Street 395, rue Wellington Ottawa ON K 1 A ON4 Otiawa ON K1A ON4 Canada Canada

The author has granted a non- exclusive licence aiiowing the National Library of Canada to reproduce, loan, distribute or sel1 copies of this thesis in microforni, paper or electronic formats.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/film, de reproduction sur papier ou sur format électronique.

The author retains ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fiom it Ni la thèse ni des extraits substantiels may be printed or otherwise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

Separate proliferation kinetics of FGF-responsive and EGF-responsive neural

stem cells within the embryonic forebrain germinal zone

A thesis submitted in conformity with the requirements for the Degree of Master

of Science, Graduate Department of Anatomy and Cell Biology, University of

Toronto, 1999

ABSTRACT

The embryonic forebrain gemiinal zone contains two separate and additive

populations of stem cells that both exhibit self-renewal and multipotentiality . While

cumulative S-phase labeling studies have investigated the proliferation kinetics of the

overall population of precunor cells within the forebrain germinal zone through brain

development, linle is known about when and how (symmetrically or asymmetri'cally) the

small subpopulations of stem cells are proliferating in vivo. This has been determined by

injecting timed-pregnant mice with high doses of tritiated thymidine ()H-T~R) to kill any

stem cells proliferating within the striatal germinal zcne in vivo and then by assaying for

neurosphere formation in vitro. Injections of 0.8 mCi of 3 ~ - ~ d ~ given every 2 h for 12 h

to timed-pregnant mice at EI 1, E14, and El7 resulted in signifiant depletions in the

nurnber of neurospheres generated by FGF-responsive stem cells at E 1 1, and by EGF-

responsive and FGF-responsive stem cells at El4 and E17. With increasing embryonic

age, the depletions observeci in the number of neurospheres generated in vitro in response

to FGF2 following exposure to 'H-T& in vivo decreasecî, suggesting there is an increase

in the length of the ceU cycle o f FGF-responsive neural stem cells through embryonic

development. The results suggest that the FGF-responsive stem ceIl population expands

between E l 1 and El4 by dividing symmetrically, but switches to prirnarily asymmeûic

division between E 14 and E 17. The EGF-responsive stem cells anse after E 1 1, and their

population expands through syrnmetric divisions and through asymmetric divisions of

FGF-responsive stem cells.

iii

Acknowtedgemen ts

First 1 would like to thank my supervisor, Dr. Derek van der Kooy (the Drek),

and the members of my thesis cornittee for their time and advice: Dr. Jane Aubin, Dr.

Mike Wiley, Dr. Annelise Jorgensen, Dr. Charles Tator and Dr. Norman Iscove.

I would not be here now if it wasn't for the constant support of my family: mom,

dad, Will, Bruce, Christy, Brendan, Kyle, L a i e and Tyson. All the e-mails, telephone

calls, and trips home helped me to keep my eye on the goal and not give up when things

got a little rough. Of course, the cash helped a lot too. I dedicate this thesis to my

nephews Brendan, Kyle and Tyson. Thinking of you three has always helped me put

things in perspective, and made me realize what is really important, family.

Last, but not least, my fellow laboids (past and present): Catherine, Cindi,

Colleen, Raewyn, Bernie, Brenda, Vince and Nirit, Glenn, Steve, Tania, Sue, Bill, my diva

Danka and especially Hance (xoxo). Where would 1 be without al1 of you bugging my ass

day in and day out. Seriously, though, thank you to Catherine, for helping me corne out

of it dl; Cindi, for my therapy sessions; Colleen, for our walks around the front campus

prior to o u defenses; Glenn, for drivhg me around; Vince, for doing al1 my expenments;

Raewyn, for being my queen; Sue, for al1 our afternoon breaks; Hance for al1 his phobias;

Steve and Brenda for sharing my space; Nirit, for king Nirit; Bill, for your knowledge of

DNA; and Tania, for doing my nails. I haven't mentioned Cindi enough. Thank you,

thank you, thank you!!! That was for al1 your advice and letting me look after your

amazing children. You have done su much for me that 1 don? know how 1 can ever repay

you.

Table of Contents

A bstract

Acknowledgements

Figure Index

Introduction

Materials and Methods

Results

Discussion

Page

i i

iv

v i

References

In adult marnrnals, a number of tissues display a tremendous ability to replace

cells that are continually lost throughout life. The cells are replaced by a small population

of undifferentiated cells known as stem cells. These cells are defined by their ability to:

(1) proliferate; (2) self-renew, maintain themselves throughout the life of the organism; (3)

produce a large number of differentiated progeny through a population of proliferating

progenitors; (4) maintain their multilineage potential over time; and (5) regencrate

damageci or diseased tissues (Hall and Watt, 1989; Potten and Loeffler, 1990). Examples

of tissues that contain populations of stem cells are the hematopoietic system, the gut

epitheliurn, and the skin epitheliurn. Up until the early nineties there was little evidence

for the existence of a stem ce11 in the mammalian brain. Indeed, the adult mammalian brain

does not e h b i t much cellular proliferation, and it is not capable of generating al1 the ce11

types necessary for regeneration to occur following injury or disease. Furthemore, during

mammalian deveiopment, al1 the neurons, astrocytes and oligodendrocytes that will

eventually make up the adult central nervous system are generated From the germinal zone

lining the ventricular neuraxis, which was thought to consist of separate neuronal and glial

progenitors (Frederikson and McKay, 1988; Skoff, 1990). AAer all, neurons are

primarily generated prenatally h i l e glia (astrocytes and oligodendrocytes) are primarily

generated postnatally. Even clonal analysis suggested neurons and glia had separate

progenitors (Luskin et al., 1988; Luskin et al., 1993; Luskin and McDermott, 1994). I t

was only within the past decade that in vitro studies have s h o w that the forebrain

, germinal zone and its adult remnant, the subependyma, contain a small population of ."-

separate EGF-responsive and FGF-responsive neural stem cells (Reynolds et al., 1992;

Reynolds and Weiss, 1992; Morshead et al., 1994; Gritti et al., 1996; Tropepe et al.,

1999), which display what is considered to be the definitive characteristics of stem cells,

self-renewal and multipotentiality (Hall and Watt, 1989; Potten and Loeffler, 1990;

Reynolds and Weiss, 1996).

Cells that displayed these characteristics of stem cells were first isolated in vitro

from the embryonic striatal genninal zone (Reynolds et al., 1992; Gritti et al., 1996), and

subsequently fiom the adult remnant of the striatal germinal zone, the subependyma

which lines the lateral ventricles (Reynolds and Weiss, 1992; Morshead et al., 1994).

Clonal and population analysis confinned the identity of these cells as stem cells

(Reynolds and Weiss, 1996; Gritti et al., 1996). Single cells isolated in vitro in the

presence of EGF or FGF2 gave rise to clonal aggregates of cells called neurospheres,

which could be used to display the self-renewal and multipotentiality of the stem cells.

This suggested that there was one neural stem ce11 population that could proliferate to

form neurospheres in response to either EGF or FGF2. However, stem cells isolated

from the embryonic striatal germinal zone at ernbryonic day (E) 14 and cultured in the

presence of both growth factors yielded additive numbers of neurospheres, as would be

expected by summing the numben of neurospheres generated in culture with either

growth factor alone (Tropepe et al., 1999). This suggests that there are two populations

I

i l. of stem cells residing in the germinal zone, defined on the basis of their growth factor

responsiveness.

At present, there is no marker that cm be used to distinguish stem cells from their

more differentiated progeny. The only tool available to study neural stem cells is the in

vitro neurosphere assay. This assay has been extremely useful in studying how the

fundamental properties of stem cells are regdateci within the Central Nervous System. In

my thesis, 1 will show that this assay can also be used, in combination with in vivo

manipulations, to gain further insight into how neural stem cells behave in vivo. There are

two pieces of evidence that support the idea of using the neurosphere assay to

investigate how stem cells behave in vivo. (1) The stem celis are cultured using optimal

growth factor concentrations, suggesting that the culture system is suficient to stimulate

al1 stem cells isolated in vitro to proliferate to form neurospheres (Tropepe et al., 1999).

(2) Over 90 % of the neurospheres generated in vitro ddisplay self-renewal and

multipotentiality (Reynolds and Weiss, 1996; Gritti et al., 1996). Therefore, the in vitro

neurosphere assay is a hghly stringent assay for studying the propenies of neural stem

cells.

Indeed, the proliferation kinetics of stem cells residing in the subependyma of the

adult mammalian brain were revealed using this assay, and in vivo clonal analysis using a

repiicationdeficient retrovirus (Moahead et al., 1994; Monhead et al., 1998). In the

adult subependyma, the FGF-responsive and EGF-responsive stem cells exist in a

, relatively quiescent state, proliferating asymmetrically approximately every 15 days to , 5 1-

generate one of itself and a constitutively proliferating progenitor ceIl, which has a œll

cycle time of 12.7 hours (Monhead and van der Kooy, 1992). The entire population of

stem cells makes up approximately 0.2-0.4 % of al1 subependymal cells and maintains the

size of its population by asymmetric division throughout the life of the rnouse (Tropepe

et al., 1997). While the proliferation kinetics of neural stem cells have been investigated in

the adult mouse brain (Monhead et al., 1994 and 1998), the fundamental questions

conceming how these separate populations of neural stem cells are established and how

they proliferate during forebrain development in vivo remains mostly unknown.

FGF-responsive neural stem cells can be isolated in vitro fiorn the developing

forebrain as early as E8.5 (Tropepe et al., 1999). The numben of isolated stem cells that

proliferate in response to FGF2 increase with increasing embryonic age (Tropepe et al.,

1999). This suggests that the stem cell is proliferating syrnrnetrically to expand its

population in vivo. The EGF-responsive stem cells are thought to arke fiom the earlier

bom FGF-responsive stem cells and appear between E 1 1 and E 13 (Tropepe et al., 1999).

In my thesis 1 sought to determine when these two populations of stem cells are

proliferating symmetncally (to expand and establish their populations), and when they

switch to primarily asymrnetncal division (to produce separate progenitor cells), whch is

how they proliferate in the adult mouse forebrain.

i 4 Using cumulative S-phase labeling with 'H-TÇIR or bromodeoxyuridine (BrdU),

the average proliferation kinetics have been estirnated for the populations of cells that

make up the germinal zone of the embryonic striatum (Bhide, 1996) and cortex (Waechter

and Jaensch, 1972; Takahashi et al., 1993; Reznikov et al., 1995; Takahashi et al., 1995).

The neural stem ce11 populations within these primordial zones are very small, less than

1% of al1 genninal zone cells (Tropepe et al., 1999). Conventional cumulative labeling

techniques, which study population averages, would not be able to detect the presence of

such a small subpopulation of stem cells within the larger precursor population, even if

the kinetics of the stem cells differed fiom their progeny. To reveal the proliferation

kinetics of stem cells 1 used high doses of 'H-T~R to kill off proliferating cells in vivo and

then isolated cells fiom the striata1 geminal zone to see if there were any stem cells lefi to

fonn neurospheres in vitro. Exposure of cells to high doses of 'H-T~R of high specific

activity is known to luIl proliferating cells by intranuclear radiation (Painter et al., 1958;

Drew and Painter, 1959; Becker et al., 1965; Lajtha et al., 1969; Morshead et al., 1994 and

1998). Accordingly, if the stem cells were proliferating at the time of exposure to 'H-

TdR, then they would incorporate the nucleotide into their DNA, die and fail to fom

neurospheres in vitro. My data show that the length of the ce11 cycle of FGF-responsive

stem cells increases between El l and El 7. Between El 1 and El4 the FGF-responsive

stem cells proliferate rapidly and symmetrically to increase the size of their population,

but then these FGF-responsive stem cells switch to prirnarily asymmetric division

between El4 and E 17. Between E 1 1 and E14, an EGF-responsive stem ce11 ernerges and

possesses a relatively long ce11 cycle time. This population then expands during

, i /. embryogenesis by symmetnc division and by asymmetric division of the FGF-responsive a--

stem cell.

MATERIALS AND METHODS

'8-T~R kill paradigms

Timed-pregnant CD I mice (Charles River, St. Constant, Quebec) at either 1 1, 14,

or 17 days of gestation were divided into two groups, one that was injected with saline

and one that was injected with 'H-T~R (Specific Activity = 50 Cihmol, 1.0 mCihL;

ICN). The animals were injected intraperitoneally with either 0.8 mCi 'H-TW or an

equivalent volume (0.8 mL) of the saline vehicle every two houn for 12 houn. In one

experiment timed pregnant mice at 14 days of gestation were injected with 0.8 mL saline

vehicle or 0.8 mCi of 3 ~ - ~ d ~ every 2 hours for 20 hours.

Dissection and in Vino neurosphere asuy

Two hours afier the last injection the mice were sacrificed by cexvical dislocation.

In contrast with analyses using conventional cwnulative labeling in vivo (during which a

group of animals is sacrificed after each injection), animals that were injected with high

doses of 'H-T~R at each age were sacrificed only afier receiving several injections due to

the high cost of using such high doses of 'H-TW. For the neurosphere assay, four

embryos were removed From each dam, two from each uterine hom. At E 11, El4 and El 7

the striatal primordia were dissected from each embryo in Dulbecco's-PBS (Gibco). At

El 1 these dissections included the striatal and pailidal ventricular zones, while at later

ages the dissections included the striatal and pallidal ventricular and subventncular zones

F i 1 ) Throughout the rest of the paper the dissected tissue will be refend to as the

& /- stnatal germinal zone. The tissue from each embryo was transferred to serum-free media m.

and triturated into singIe cells. Cells were cultured in duplicate at a density of 50 celldpL

in 24 well plates containing chemically defined SFM (DMEMIF 12 (1: 1 ) (Life

Technologies, Gaithersburg, MD), 5 m M HEPES buffer, 0.6% glucose, 3 mM NaHCO,,

2 rnM glutamine, 25 p g h L insu1 in, 1 00 pghL transfemn, 20 nM progesterone, 60 p M

putrescine, and 30 nM sodium selenite) in the presence of either EGF (20 ng/mL; purified

from mouse submaxillary gland; Upstate Biotechnology, Lake Placid, NY) or FGF2 (10

ng1m.L; human recombinant; Upstate Biotechnology, Lake Placid, NY) and heparin (2

pgfmL; Sigma). To determine if any of the decrease obsewed in the number of

neurospheres that were generated in vitro was the result of stem cells incorporating the

'H-T~R in vitro fiom dying cells, cells fiom each embryo were cultured in duplicate in

defined growth factor supplemented media containing cold thymidine (Sigma) and 2'-

deoxycytidine (Sigma) at concentrations of 100 pg/rnL and 20 pg/mL, respectively. The

total number of neurospheres that fomed in each well was counted afkr six days in vitro

(DIV). Neurospheres were counted &ter six DIV since al1 the neurospheres that will fomi

have fomed by this time (Monhead et al., 1998). Although the number of neurospheres

does not increase after six DIV, the ones that have formed continue to increase in size.

The data are reported as the mean number of neurospheres (2 SEM) fomed from one

dam, which is the average of the results from four embryos per dam. In order to

determine that the neurospheres generated fiom striatal germinal zone tissue in al1 of the

groups were derived from stem cells, I assayed for self-renewal and multipotentiality as

previously described (Tropepe et al., 1999).

Figure 1. Coronal sections through the forebrain of the mouse at E 1 1, E 14 and El 7

(Altman and Bayer, 1995). The areas within each box depict what tissues were included

in the striatal germinal zone dissections at each embryonic age. At El 1, the dissections

included the striatal and pallidal ventricuiar zones, whereas at both El4 and El 7 the

dissections included the striatal and pallidal ventricular and subventricular zones

(separated from each other by a white line). C - cortical germinal zone; S - striatal

germinal zone; P - pallidal germinal zone; V - ventncle.

Incorporation of th^ into DNA

lmmediately following the 12 houn of injections at either E 1 1, E 14, or El 7 striata

were dissected frorn four embryoddam and frozen at -70°C. For the determination of the

arnount of 'H-T~R incorporated into the DNA of proliferating cells, striata were

incubated in an extraction buffer (50 m M Tris-HC1, pH 8.0; 100 mM EDTA; 100 mM

NaCI; 1% SDS; 0.5 mg/rnL Proteinau: K (14mg/mL, Boehringer Mannheim)) ovemight at

55°C. Protein and lipids were removed by the addition of saturated NaCl (4 M). DNA

was precipitated with an equal volume of isopropanol and the resultant pellet was

washed with 70% ethanol. The pellet was air dried and resuspended in Tris-EDTA (TE)

buffer (pH 8.0). The DNA concentration was determined measuring the absorption of

W Iight at 260 nm on a Beckman DU-8B spectrophotometer. For scintillation counting

duplicate samples were prepared by pipetting 100 pL samples of the DNA suspension

ont0 2.4 cm glass fiber filter circles (Fisher scientific). Afler drying, the DNA was

precipitated ont0 the filter circles by m i n g them through the following solutions, ten

minutes in each solution: 10% TCA, 5% TCA (both dissolved in 0.1 M PBS), 95%

ethanol, and 95% ethanol. The filter circles were allowed to air dry completely before

putting them in glass vials containing 10 mL of scintillation fluid (Cytoscint). AAer

sitting for one hour, the number of scintillations per minute (cpm) for each sample was

deterrnined using a Bechan LS 1800 scintillation counter. Background radiation counts

were determined using the sarne protocol except LOO pL of TE were pippetted onio the

glas fiber filter circles. The specific radioactivity was reported as the mean + SEM

#

cpdpg DNA per dam, which was the average fiom the dissections of four embryos per

dam.

Estimates of stem ceil num bers

The number of stem cells residing within the striatal gemünal zone was estimated

from the results of the neurosphere assay of embyos from saline injected dams at El 1,

El4 and El 7. The total number of viable cells per striatal germinal zone dissection was

determined by counting a sample of the ceIl suspension from each embryo using the

Trypan blue exclusion method. Other cells fiom the suspension were cdtured at a

density of 50 celldpL, giving a total of 25 000 viable cells/well. Estimates of the total

number of stem cells residing in the srnatal germinal zone at El 1, El4 and El7 were

determined by first dividing the average number of neurospheres generated fiom four

embryos by the number of viable cells plated per well. This number was then multiplieci

by the total number of viable cells per striatal genninal zone dissection, which was the

average nurnber of viable cells fiom the dissections of four embryos per dam.

RESULTS

Incorporation of 3 ~ - ~ d ~ in vivo decreases the num ber of neurospheres tbat form

in vr'rro

At El 1, stem cells isolated from the striatal gexminal zone generate clona!

aggregates (neurospheres) only in the presence of FGF2. Later in embryogenesis striatal

gemind zone stem cells generate neurospheres in response to EGF. There is evidence

that separate and additive populations of neural stem cells fom neurospheres in response

to EGF and FGF2 at El4 and that the EGF-responsive stem cells are the progeny of

FGF-responsive stem cells (Tropepe et al., 1999). Chimetic rnice with nul1 mutations of

the FGF receptor 1 in some of the striata1 germinal zone cells faii to develop either FGF-

responsive neural stem cells at E 1 l or separate EGF-responsive neural stem cells at El4

from the mutant cells, although both types of neural stem cells develop from the wild

type striatal germinal zone cells from the same mice (Tropepe et al., 1999). Therefore, at

E 14 and E 1 7 the effect of 3 ~ - ~ d ~ on the number of neurospheres that fonn in vitro was

determined separately for EGF-responsive and FGF-responsive stem cells.

At al1 three ages dams were injected with 0.8 mCi of 'H-T~R every 2 hours for 12

hours. Greater depletions of FGF-responsive stem cells were seen at earlier embryonic

ages (Fig. 2). For FGF-responsive stem cells, an ANOVA comparing the relative nurnber

of neurospheres generated across al1 three ages from saline treated or 'H-T~R treated

dams showed a significant effect of dnig treatment (F1,117=47.53, p<0.05), a significant

, effect of age (F2,, 17=4.36, pc0.05), and a signifiant h g w age interaction (F2,, 17==4.36,

pd.05). Incorporation of the 3 ~ - ~ d ~ into proliferating striatal germinal zone cells at

El l , E14, and El7 resulted, respectively, in approxirnately 88%, 60% and 27%

depletions of FGF-responsive stem cells compared to saline treated controls (Fig. 2).

Injections of 'H-T~R depleted the number of neurospheres generated in FGF2 at El 1

significantly more than at either El4 (tdO=2.86, p<0.05) or El7 (t&.79, p<0.05). With

increasing embryonic age, there was a significant decrease in the percent of stem cells

depleted following exposure to 3 ~ - ~ d ~ , suggesting there was a change in the kinetics of

the FGF-responsive stem cells over time.

Incorporation of 3 ~ - ~ d ~ into the proliferating striatal gemiinal zone cells at both

El4 and El7 resulted in a depletion of approximately 35-40% of EGF-responsive stem

cells compared to saline treated controls (Fig. 2). An ANOVA cornparhg the relative

numbers of EGF-denved neurospheres generated at E 14 and E 17 from saline treated or

3H-TdR treated dams showed a significant overall effect of 3 ~ - ~ d ~ treatment

(F,,9,=6.362, p <O.OS), but no effect of age (p>O.O5) and no interaction between the two

(p>0.05). There was no difference observed in the percent depletion of EGF-responsive

stem cells between El4 and E17, suggesting the kinetics of this separate population of

neural stem cells do not change over this time.

Proportions of the stem cells at both El4 and El7 (40-73%) remain, following

exposure to 'H-T~R for 12 hours. Assuming single population kinetics for FGF-

8

& /. responsive stem cells, it can be hypothesized that the length of the cell cycle of FGF- . .

responsive stem cells increased between El 1 and E 14 and again between E 14 and E 17.

This hypothesis predicts that injecting 'H-T~R for longer than 12 hours at either El4 or

El7 might deplete funher the number of neurospheres generated in vitro in response to

EGF or FGF2. Altematively, if there are two separate cycling populations within each of

the EGF- and FGF-responsive stem cell populations at El4 or E17, then the stem ceils

labeled within 12 hours represent a population with a shoner cell cycle time, while those

that don't incorporate 'H-T~R during the 12 hours may represent a population of cells

with a longer ce11 cycle time (Waechter and Jaensch, 1972; Nowakowski et al., 1989).

The two population assumption would again predict that injecting 'H-T~R for longer

than 12 hours at either age would further deplete the number of neurospheres generated in

vitro in response to EGF or FGF2, by depleting more of the stem cells that have longer

ce11 cycle times.

'H-T~R release in vitro does not account for the depletion of neurospheres

The decrease in the relative numbers of neurospheres depleted compared to saline

controls suggests that the stem cells were proliferating at the tirne of exposure to the

nucleotide in vivo. This conclusion is based on the lack of neurosphere formation in vitro.

It is also possible that some of the effects of 'H-T~R could have ken the result of in

vitro nucleotide release from dying cells into the wells of the culture dish, the subsequent

incorporation of 'H-T& into proliferating stem cells in vitro, and then stem ceIl death as

the cells attempt to undergo mitosis and form neurospheres in vitro. This latter result

Figure 2. Cornparison of the percent of neurospheres fomed in either EGF or FGF2

following incorporation of 'H-T~R at E 1 1, E 14, and El 7. The 1 00% saline control value

for each age group is not show. At El 1 , neurospheres do not form when cells are

cultured in EGF. Data represent means + SEM. (El 1, n=12 dams (average counts fiom

four embryos per dam); E14, n=30 dams; E17, n=20 dams).

Number of neurospheres as a percent of saline injected controls

0

would decrease the numbers of neurospheres that form after culturing the cells for 6 days .$'-

in vitro, but the decrease would be misinterpreted as king representative of how many

stem cells were proliferating in vivo. To nile out this in vitro kill explanation, dissociated

striatal cells from embryos treated in vivo with 'H-T~R or saline vehicle were cultured in

the presence of high doses of nonradioactive (cold) thymidine and 2'-deoxycytidine

(Hasthorpe and Harris, 1979). High doses of cold thymidine will compte with the 3 ~ -

TdR for incorporation into the DNA of cells in S phase. The high dose of cold thymidine

needed to prevent incorporation of 'H-T~R into the DNA is cytotoxic in and of itself

unless 2'-deoxycytidine is also added to the culture medium (Bjursell and Reichard,

1973). After 'H-T~R treatment at El1 the depletions in the absolute number of

neurospheres generated in the presence of cold thymidine were the same as those

observed in the numbers of neurospheres generated in its absence (Fig. 3). An ANOVA

comparing the absolute numbers of neurospheres generated h m dams injected with saline

or th th^ at E 1 1 and cultured in the presence or absence of cold thymidine and 2'-

deoxycytidine showed a significant effect of in vivo drug treatment on the formation of

neurospheres (FIt8=14.46, p<O.OS), but no effect of the absence or presence of cold

thymidine, nor any interaction. Depletions in neurospheres by 'H-T~R in vivo also were

similar in the presence or absence of cold thymidine and 2'4eoxycytidine when dams

were injected with 'H-T~R at El4 and E17. Therefore, 1 conclude that the decrease in the

number of neurospheres generated m vitro from striata at E 1 1, E 14, and E 17 was a result

of neural stem cells dying after incorporating 'H-T~R in vivo.

Figure 3. The number of neurospheres culhired at El 1 in the absence or presence of cold

thymidine and Zt-ûeoxycytidine. Culturing striatal germinal zone cells in the presence of

unlabeled nucleosides does not affect the depletion observeci in the number of

neurospheres that form in vitro following in vivo incorporation of high doses of 'H-T~R.

Data represent the means + SEM of the numben of neurospheres generated per well (25

000 viable cells per well) from saline (n=9 dams) and 'FI-thy (n= 12 dams) treated dams

(rnean of four embryos per dam).

Saline

FGF2 FGF2+Cold TdR

6 r. I t is also possible that the nucleotide in vitro may have produced more subtle ..

effecis on the growth of neurospheres, their passaging, and their differentiation potential.

High doses of thymidine have ken observed to inhibit the completion of S-phase and

hence of mitosis (Bjursell and Reichard, 1973). This hypothesis might imply that the

depletion in the number of neurospheres in vitro may be a result of a delay in neurosphere

formation. To test this possibility neurospheres were counted after nine days in vitro

(rather than the six days initially employed), after which we might see the full

complement of neurospheres from 3 ~ - ~ d ~ injected animals compared to those injected

with saline. At E 1 1, following exposure to 'H-T~R, the numben of neurospheres were

counted at six days in vitro and again after nine days in vitro. There was no difference

observed between the number of neurospheres counted afler six days compared to afier

nine days from saline treated ( Z 6 days563.8~ 16.4 1, ii 9 days=63.8~17.17; t4=0,

p>0.05) or 'H-T~R treated ( F6 days=7.5~3.76, E 9 day~7.223.93; t6=0.06, p>0.05)

dams.

To determine if the stem cells remaining after exposure to 'H-T~R at E 1 1, El 4 and

E 17 still retained their characteristics of self-renewal and multipotentiality, neurospheres

generated at each age from both saline injected animals and 'H-T~R injected animals were

passaged and processed for imrnunohistochemistry with cell-speci fic antibodies (anti-

MAP2 for neurons, anti-GFAP for astrocytes, anti-04 for oligodendrocytes). Single

neurospheres from saline or 3 ~ - ~ d ~ treated animals, generated fiom each of the El 1, E 14,

, and El7 striatal germinal zones, were successfully passaged for three generations in EGF Y-

or FGF?. Also, single neurospheres, generateâ from El I , E14, and El 7 striatal germinal

zones f ie r saline or 'H-T~R treatment in vivo and then differentiated in vitro, contained

neurons, astrocytes, and oligodendrocytes. Therefore, injecting pregnant mice with high

doses of 'H-T~R does not block the ability of the individual remaining neurospheres

generated in vitro to be passaged or for their cells to differentiate into neurons, astrocytes

and oligodendrocytes in vitro.

The incorporation of 'H-T~R into DNA of striatal germinal zone cells i s less at

El7 than at El1

For FGF-responsive stem cells the relative numbers of neurospheres depleted at

El 1, El 4 or El 7 (compared to controls at the sarne aga) decreased with increasing

embryonic age. There are two possibilities that could explain these results: (1) the dose

of 'H-T~R used wûs insufficient to kill the cells that had incorporated the nucleotide

throughout its administration at the older ages, and (2) the kinetics of the stem cells have

changed (see below). The first possibility was tested by detennining the amount of 'H-

TdR incorporated into the DNA of striata1 germinal zone cells at E l l, E 14 and E 17. Less

'H-T~R was incorporated per ug DNA at El7 than at the earlier embryonic ages (Fig. 4).

An ANOVA cornparhg the amounts of 'H-T~R incorporated per pg DNA extracted from

striatal germinal zone cells following saline or 3 ~ - ~ d ~ treatment at E 11, El4 and E 17

showed a signifiant effect of dmg (F 1,70=77.60, p<O.O5), age (F1,70=6.98, ~€0.05) and a

signifiant h g x age interaction (F2,70=6.76, p4 .05) . There was no significant difference

, in the radioactivity incorporated into the DNA at El4 compared to El l (t2,=1.05, .IL

~ ~ 0 . 0 5 ) . However, the amount of 3 ~ - ~ d ~ incorporated into the DNA of cells within the

striatal germinal zone was significantly less at El7 than at El l (tZ2=4.93, p<0.05) or at

El4 (t3*=3.0 1, p<0.05). Based on these results, it remains unclear whether the decrease in

the depletion of neurospheres at El7 by 3 ~ - ~ d ~ treatment was a function of increasing

ce11 cycle length with embryonic age, or whether at El7 the dose of 'H-T~R was not

suffiicient to kill cells that incorporated the nucleotide.

Increasing the dose of 'H-T~R at El7 does not furtber deplete the nurnber of

neurospheres formed in vitro or increase the incorporation of 3 ~ - ~ d ~ i n vivo

The first possibility for the decreased depletion of neural stem cdls frorn the

striatal pnninal zone by ~H-TW treatment with increasing ernbryonic age was further

exarnined by injecting animals with a higher dose of 'H-T~R at E17. A dose of 0.8 mCi of

'H-T~R injected evety 2 hours for 12 hours was sufXcient to kill off the entire

constitutively proliferating population of cells, 50% of the stem cells lining the lateral

ventricle of the adult mouse brain (Monhead et al., 1994), and clearly can kill stem cells

proliferating at earlier embryonic ages (the present results). At E 1 l this dose, injected

evexy 2 hours for 12 hours, killed up to 90 % of stem cells residing in the striatal germinal

mne. If the length of the ceil cycle of the stem ceU does not change over embryonic time

(the second explanation), then our alternative (first) explanation for the decreased

depletion of stem cells at El7 compared to El l suggests that the dose of 3 ~ - ~ d R injected

over 12 h o w is not suficient to hl1 off the proliferating cells. Therefore, &mals at El7

Figure 4. The amount of 'H-T& incorporated per pg DNA in striatal germinal zone ceUs

at El 1 , E14, and El7 following 12 hours of injections of 'H-T~R. The specific

l radioactivity of DNA is significantly lower at El7 than at either E l 1 or E14. Data

represent means + SEM of the average cpm/pg DNA deterrnined fiom bilateral striatal

germinal zone dissections of four embryos per saline or 'H-T~R treated dam at El 1 (n=8

dams), E 14 (n= 16 dams), and E 17 (n= 16 dams).

were injected once every 2 hours for 12 hours with either 0.8 mCi of 3 ~ - ~ d ~ per .cX

injection (low dose) or 1.6 mCi of 3 ~ - ~ d ~ per injection (hgh dose) to test whether

increasing the dose of each injection over a 12 hour period would result in a greater

depletion in neurosphere formation. No greater depletion of neurospheres was seen with

the higher 'H-T~R dose compared to the lower 'H-T~R dose (Fig. 5A). An ANOVA

companng the absolute nmben of neurospheres generated in EGF or FGF2 following

saline treatment or injections of low or high doses of ~ H - T ~ R at El 7 showed no effect of

growth factor conditions (FlVl8=1.80, p>O.OS), a significant main effect of 'H-T~R

treatment (F2,18=8.26, p>O.OS), but no significant interaction (F2,i8=0.î0, p>O.OS).

Although there was an effect of 'H-T~R cornpared to its saline vehicle on the absolute

numben of neurospheres generated, the neural stem cells were equally depleted in the low

dose and hi& dose of 'H-T~R groups (EGF, t6=0.S 1, ps0.05; FGF, t 4 . 6 7 , p>0.05).

Furthemore, no difference was observeci between the low dose group and the high

dose group in the amount of 'H-T~R incorporated into the DNA of cells dissected from

the striatal germinal zone at E17 (t&.55, p>0.05) (Fig. SB). Given that no further

decrease in the numbers of neurospheres was seen after injecting animals with a higher

dose of ~H-T~R and no further increase was seen in 'H-T~R incorporation per ce11 within

the germinal zone at E17, I suggest that the 'H-T~R uptake into the DNA of al1 cells in S-

phase over the entire injection period was saturated. Thus, I conclude that the decrease in

neural stem cells killed by 'H-T~R at El7 compared to El l reflects the longer ce11 cycle

time of stem cells at El 7. The lower 3 ~ - ~ d ~ incorporation at El7 than at El 1 does not

, reflect the lack of saturation of the labelhg of the DNA at E17, but rather the fact that .*-

either fewer cells at El7 were in S-phase at the time of exposure to 'H-T~R or the El 7

dissections included more of the postmitotic striatum than the earlier embryonic

dissections.

High doses of 'H-T~R do not alter the kioetics of the remaining stem cells

proliferating in vivo

Given that the dose of 'H-T~R per injection was suffi~cient to kill ali the cells

proliferating within 12 houn at each age, the decrease in the relative numben of FGF-

responsive stem cells depleted compared to saline controls with increasing embryonic age

was most likely the result of a change in the kinetics of the cells. Two types of changes

are possible: (1) the stem cells at the later embryonic ages changeci their kinetics to

replenish their populations before the cells were cultured to geneiate neurospheres in

vitro, and (2) the length of the ceIl cycle of the stem cells increased with increasing

embryonic age (see below). The first possibility seems unlikely given the 88% depletion

in the numbers of neurospheres generated following 'H-T~R treatment at El 1 compared

to saline treated controls. Moreover, even with long pst-kill survival times in vivo, the

stem ce11 populations did not recover to saline vehicle baseline leveis when animals were

injected at El4 and their striata were dissected out at postrnitotic day (P) 56. The

numben of neurospheres generated frorn addt mice in either EGF or FGFZ following 'H-

TdR treatment at El4 were significantly less than that of addt mice that had been treated

with saline at El4 (EGF, t6=3.45, pXO.05; FGF, t8=2.88, p<0.05). At P56, the El4 3 ~ -

Figure 5 A. The number of neurospheres generated in EGF or FGF2 following injections

of saline, 0.8 mCi/injection of 3 ~ - ~ d ~ , or 1.6 mWinjection of 3 ~ - ~ d ~ at E 17. Injections

of a higher dose of 'H-T~R does not further decrease the number of neurospheres that

1 fonn compared to the lower dose of 'H-T~R. Data represent means 2 SEM of the

numbers of neurospheres generated per well (25 000 viable cells/well) from saline (n=4

dams), low dose 3 ~ - ~ d ~ (n=4 dams), and high dose 'H-T~R (n=4 dams) treated dams

( m m of four embryos per dam).

B. The specific radioactivity of the DNA extracted from striatal germinal zone cells at

E 1 7 following injections of 0.8 mCi/injection of 'H-T~R and 1 .6 mcilinjection of 3 ~ - ~ d ~ .

Injections of the higher dose of 'H-T~R does not further increase the incorporation of the

nucleotide into the DNA of dividing germinal zone cells compared to the incorporation at

the lower 3 ~ - ~ d ~ dose. Data represents means + SEM of the average cpm/pg DNA

detemined from bilateral striatal germinai zone dissections of four ernbryos from each of

the low dose 'H-T~R (n=4 dams) and high dose -'H-T& (n=4 dams) treated dams.

cpmtug DNA Number of neurospheres -

per well

TdR treated mice were srnaller than their saline treated controls, in both body and brain .ii

weights, and the percent depletion observed in the nwnbers of neurospheres generateà in

EGF and FGF2 from P56 mice exposed to 'H-T~R at El4 was not signifimtly different

fiom that observed after sacrifice at El4 (p>0.05). It is likely then that the remaining

stem cells afler 'H-T~R treatment at El4 were not capable of replenishing their

population in vivo through symmetnc division. Similar results have been observed

following two series of high dose injections of 'H-T~R in the subependyma of the adult

mouse forebrain (Morshead et al., 1994). When a series of 'H-T~R treatment was given to

adult animals to deplete the constitutively proliferating population (Tc = 12.7 hours),

stem cells were recruited to divide to repopulate the lost progeny. During the recnùtment

phase, a second senes of injections was then given to the same mice which depleted the

proliferating stem cells by 50% (Monhead et al., 1994). Neither the neural stem ceIl

poulation nor its progeny (the constitutively proliferating population) were replenished

to control values afier 50% of the stem cells were killed with the second series of 3 ~ - ~ d ~

injections (Morshead et al., 1994).

Exposure to 'H-T~R for a longer period of time at El4 further depletes the number

of FGF-responsive stem cells that give rise to neurospheres in vitro

Given that the stem cells were unable to change their kinetics to replenish their

population following 3 ~ - ~ d ~ treatment at E14, 1 hypothesized that the decreased

depletion in neurospheres with increasing embryonic age was a result of the length of the

ce11 cycle increasing with age. According to the method of cumulative S-phase labeling,

injections of 'H-T~R over a longer penod of time should result in a greater depletion of .ii

neurospheres as more and more of the asynchronously proliferating neural stem cells

incorporate the nucleotide and are killed [at least until the injections have covered the

lengh of the entire ce11 cycle (Tc) minus the length of S-phase (T,)]. At E 14, injections of

'H-T~R every 2 hours for 12 hours resulted in a 60% depletion in the number of

neurospheres that forrned in vitro in response to FGFZ. Assuming that the 60%

depletion of neurospheres accounted for 60% of Tc-Ts, 1 hypothesized that injecting 3 ~ -

TâR every 2 hours for an additional eight hours should be suff,cient to deplete the stem

ceU population to zero, if the entire stem ceIl population was cycling asynchronously

with one common ce11 cycle time (if 0.6(Tc-Ts) = 12 hours, then Tc-Ts = 20 hours).

Therefore, dams were injected at E14 with O.8mCi ' H - T ~ R every two hours for either 12

hours or 20 hours. At the end of each injection period the dams were killed and the striatal

germinal zone was dissected from four embryos from each dam. The cells were cultured

in FGFZ and the number of neurospheres were counted after six days in vitro. 'H-T~R

injections over 20 houo produced a greater depletion of neurospheres at E l 4 than was

observed afier 12 houn of 3 ~ - ~ d R (Fig. 6). The number of FGF-responsive

neurospheres generated afler 20 hours of 3 ~ - ~ d ~ injections was significantly less (a 90%

depletion) than that generated afier 12 houn of 3 ~ - ~ d ~ injections (an 80% depletion)

(t5=2.90, pe0.05). The 80% depletion observed following 12 houn of 'H-T~R treatment

seems sornewhat greater than the initial 60% depletion previously observed afier E l 4

treatments (see Fig. 2). However, figure 1 shows the results of several experiments done

Figure 6. The numbers of neurospheres generated in FGF following injections of saline or

'H-T~R for either 12 hours or 20 houn. The nurnbers of neurospheres f o n d after 20

I houn of 'H-T~R were significantly different fiom those observai after 12 hows of 'H-

TdR. Data represent means i SEM of the nurnbers of neurospheres generated per well

(25 000 viable celldwell) fiom saline (n=4 dams) and 'H-TW ( s 3 dams) treated dams at

12 hours and saline (n=4 dams) and 'H-T~R (n=4 dams) treated dams at 20 houa. One

ciam represents the mean of four embryos.

Number of neurospheres generated per well containing FGF2

#

at each embryonic age and the 80% depletion in FGF2 generated neurospheres fiom this .+'-

single experiment is indicative of the variability observed between experiments previously

done at E l 4 (and the reason for m i n g appropriate controls for each individual

expenment). Regardless of the percent depletion observed afier 'H-T~R treatment at El4

relative to saline injected controls, it is clear that a small population of stem cells remains

following 20 houn of 'H-T~R treatment, which was the predicted arnount of time the

dams needed to be exposed to 3 ~ - ~ d ~ to deplete the population of FGF-responsive stem

cells to zero.

The greater depletion with 20 hours of 'H-T~R injections cornpared to 12 hours

of injections within this expriment is unlikely to be a by-product of the higher total dose

of 'H-T~R received in the 20 hours group, given that injecting dams at El4 with 2

mCi/injection of 'H-T~R (2.5 x each dose) every 2 hours for 12 hours did not significantly

increase the relative depletions of neurospheres. However, after an additional eight hours

of 'H-T~R injections on El4 the incorporation of 3 ~ - ~ d R into the DNA of gemiinal zone

cells increased significantly from 2,109 2 33 1 cpm/pg DNA after 12 hours to 5,373 i. 552

cpdpg DNA afler 20 hours (t5=5.30, ~ 4 . 0 5 ) . Thus, more proliferating cells in vivo

were incorporating 'H-T~R after 20 hours than afler 12 hours, suggesting more of the

cells passed through S-phase within 20 hours than 12 hours. These data, in combination

with greater neurosphere depletions after 20 houn of 'H-T~R compared to 12 hours,

suggest that more of the FGF-responsive stem cells were passing through S-phase and

incorporating 3 ~ - ~ d ~ afier king exposed to the nucleotide for an additional eight hours.

, Assuming a single population of FGF-responsive stem cells al1 having the sarne cell cycle ."-

time, I conclude that the FGF-responsive stem cells have a relatively long ceIl cycle time

at E 14 compared to El 1. However, my initial hypothesis and conclusion were made on

the asswnption that with every injection of 'H-T~R there would be a proportional

decrease in stem cells, which (when plotted over tirne) would be linear. In the present

experiment the FGF-responsive stem ceU population was depleted by 80% following 12

hours of 'H-T~R treatment at E14. If I assume there is a single population of FGF-

responsive stem cells having one cornmon cell cycle time, 1 would predict that 15 houn of

'H-T~R treatment should be suffiicient to kill 100% of the stem cells (if O.g(Tc-Ts) = 12

houn, then Tc-Ts = 15 hours). Given that only 90% of the stem cells were depleted

following 20 hours of 'H-T~R treatment, 1 conclude that there must be a relatively small

second population of proliferating FGF-responsive stem cells with a very long ce11 cycle

time. If the 80% depletion in FGF-responsive stem cells following 'H-T~R treatment

represented the size of the first population, and the remaining 20% of the stem cells

represented the second population, then the length of Tc-Ts for the second population is

approximately 40 hours (i.e. it took 20 hours to deplete 50% of the second population,

therefore since O.S(Tc-Ts) = 20 hours, then Tc-Ts = 40 hours). In the adult mouse

forebrain, stem cells residing in the subependyma have a relatively long ce11 cycle time,

estimated to be at least 15 days in length (Monhead et al., 1998). Perhaps a very small

fraction of the forebrain FGF-responsive stem cells at El4 (< 10%) already have changed

their ce11 cycle time to that estimated for stem cells in the adult forebrain.

The numbers of stem cells witbin the striatal germinal zone increases through the -+''

em bryonic period

The symmetric venus asymrnetric division patterns of neural stem cells through

embryogenesis were estimated, given the absolute numbers of neurospheres generated in

vitro from the populations of striatal germinal zone cells dissected from embryos at El 1,

El 4 and E 17. These absolute estimates depend on the assumptions that ( 1 ) upon initial

dissection, ce11 viability is equally high, (2) the neurosphere assay is optimal to select for

the swival and proliferation of al1 neural stem cells residing in the striatal germinal zone

and (3) there is no ce11 death in the stem ce11 populations over time. Between E 1 1 and E 14

the number of FGF-responsive stem cells increases 12-fold (Fig. 7). This suggests that

the stem ce11 is undergoing symmetric divisions to increase the size of its population over

this three day period. I-iowever, over the three day period between E 14 and E 17 the size

of the FGF-responsive stem cell population increases by only 2-fold. There are two

possible explanations for the smaller expansion observed in the number of FGF-

responsive stem cells between E 14 and El 7 than between El 1 and E14. ( 1 ) These data

suggest that symmetric divisions within the FGF-responsive stem cell population occur

more fiequently between E l 1 and E14, than between El4 and E17, and that between El4

and El7 the mode of division has switched to more asymrnetric divisions because the

stem cell population expands only 2-fold over the later three day period compared to the

IZfold increase over the earlier three day embryonic period. (2) The FGF-responsive

Figure 7. Estimates of the total nurnbers of EGF-responsive and FGF-responsive newal

stem cells that reside in the striatal gertnuial zone at El 1, E14, and El 7. Data represent

means 2 SEM of the estimated number of stem cells residing in the srnatal germinal zone

at El 1 (n=9 dams), El4 (n=32 dams), and El 7 (n=20 dams). One dam represents the

mean of four embryos.

EGF

#

stem ceIl divides symmetncally throughout the embryonic period and the change in the .+"

expansion rate is a function of an increase in the ce11 cycle time of FGF-responsive stem

cells with increasing embryonic age. These explanations may not be mutually exclusive.

The length of the ceIl cycle may increase concurrently with a change in the mode of the

division of the FGF-responsive stem ce11 fiom symmetic to asymmetric division.

However, cornparing the E 1 1 -E 14 and E 14-El 7 penods, the increase in the ceIl cycle time

would appear to be much less than that needed to account for the changes fiom a 12-fold

to a 2-fold expansion in stem cells, suggdng that the stem cells dividing over the later

three day pied must be going through more asymmetncal than symmetrical divisions.

An EGF-responsive stem ceIl population arises between El1 and El4 and

increases in absolute numbers between E 14 and E 17 (Fig. 7). Between El4 and El7 the

EGF-responsive stem ce11 population increases by 5-fold. This suggests that this

population of EGF-responsive stem cells is also undergoing symmetnc divisions. There

is evidence that the EGF-responsive stem ce11 is the progeny of the FGF-responsive stem

cell (Tropepe et al., 1999). Therefore, the switch in mode of division Rom syrnmetric to

asymmetnc in the FGF-responsive stem ce11 population may be (in addition to generating

striatal neuronal or glial progenitors) giving rise to EGF-responsive stem cells.

Although at E 14 there is some evidence to suggea that there are two populations

of cells (with different ce11 cycle times) within the FGF-responsive stem ceIl population,

the simplest estimates of ceIl cycle times for the FGF- and EGF-responsive populations

, of neural stem cells were made assuming that each growth factor responsive stem ce11 .+'-

population consisted of single proliferating populations. Between E 1 I and E 14 the FGF-

responsive stem ceIl population expands by 12-fold in total number, suggesting that the

stem ceIl is undergoing symmetric divisions. At E I 1, 12 hours of 'H-T~R treatment

depleted the FGF-responsive stem ce11 population by 90%. To obtain the same 90%

depletion at E14, the dams were injected with 'H-T~R for 20 hours. The time it takes to

label the entire population of FGF-responsive stem cells (that is to deplete the

population to zero) quals the total ceIl cycle time (Tc) minus the time it takes to

complete S-phase (T,). Estimates of Tc-T, were made assurning single population

kinetics for FGF-responsive stem cells at each of E 1 1 and El4 (ignoring for the moment

the evidence mentioned above suggesting a small minority population of FGF-responsive

stem cells at El4 with longer ce11 cycle tirnes). Given that the 90% depletions in

neurospheres at each age represented 0.9(Tc-T,), the estimate of how long it would take

to deplete the entire population to zero could be detemined by solving for Tc-T, at each

age. If 0.9(TC-T,) = 12 hours at E 1 1, then Tc-T, = 13.2 hours, and if 0.9(Tc-T,) = 20

hours at E14, then Tc-T, = 22 hours. The total length of the ce11 cycle can be estimated

knowing T,. This ce11 cycle parameter cannot be determined €rom the results of the

present experiments. However, estimates of T, for cortical and striatal genninal zone cells

range between 4 and 6 hr and remain relatively constant through the embryonic period

(Takahashi et al., 1995; Reznikov and van der Kooy, 1995; Bhide, 19%). Assurning T, of

stem cells does not differ significantly fiom the majority of proliferating cells within the

germinal zone, the minimum estirnate for Tc = 17.2 hr at El l and 26 hr at E l 4 These

, assumptions also were used to estimate the ce11 cycle tirnes of EGF-responsive stem cells .+'-

at E 14 and E 17 and FGF-responsive stem cells at E 17 (Table 1). The values predicted for

Tc at each age may be overestimates of the actual ceil cycle time. Indeeâ, with ln vivo

cumulative labeling techniques it is necessary to take into account the length of S phase

when detexmining ce11 cycle parameten, since a single injection of the nucleotide will mark

al1 cells regardless of whether they are at the beginning or end of S phase. B y using high

doses of 'H-T~R, however, the cells exposed to the nucleoside may need a certain arnount

of time to incorporate enough of the nucleotide to be killed by intranuclear radiation.

Therefore, the time it takes to deplete the population to zero may actually be longer than

Tc-T, and thus this time would be closer to the actual (slightly shorter than estimated) ceIl

cycle time of the stem ce11 population.

Given the sizes of each stem ceIl population at each embryonic age and the length

- of their ce11 cycle times, the mode of division (symmetric ar asyrnme(n'c) can be

determined for each stem ce11 population at each embryonic age (summarized in Table 2).

At El 1 , al1 the FGF-responsive stem cells would have to divide symmetrically four times

within three days to increase the population to the number observed at El4 (a 12-fold

increase). With an estimated ce11 cycle time for FGF-responsive stem cells of 17.2 to 26

hours between E 1 1 and E 14, the stem cells would have time for a maximum of 4 divisions

each. This suggests that al1 the divisions of the FGF-responsive stem ce11 between El 1

and El4 are symmetric. However, given that striatal neurons and EGF-responsive stem

cells are also generated b e ~ e e n El 1 and E 14, at least some of the FGF-responsive stem

Table 1. Ce11 cycle times of each stem ce11 population at E l l , El4 and E17.

The length of the ce11 cycle was estimated assuming that each stem ce11

population is a single population cycling asynùuonously. Tc was estimated

using the following equation: x(Tc-Ts)=12 or 20 h, where x = the percent

depletion in the number of neurospheres following 12 or 20 hours of 3 ~ -

TIR treatment, and T, = 4 h.

L

E l 1

E l 4 "

E l 7

FGF-responsive stem cells

17.2

26

48.4

EGF-responsive stem cells

1

34

38.3

Table 2. Stem ce11 ce11 cycle times and mode of division

- -

FGF-responsive stem cells

-short ce11 cycle time

-primarily symmetric divisions

-increase in length of ce11 cycle time

-switch from symmetric to primarily asymmetric divisions

-further increase in length of ce11 cycle time

-primarily asymmetric divisions?

stem cells

-relatively long ce11 cycle time

-expands population by syrnmetric divisions and asymmetric divisions of FGF-responsive stem cells

-

-1ittle change in length of ce11 cycle time

-primarily asymmetric divisions?

, tells also must undergo asymmetric divisions. 1 presented evidence for some variability in .+

the ce11 cycle times of the FGF-responsive stem ceIl population, with the majority of

stem cells at El4 having a short cell cycle time while the rest have a relatively long ce11

cycle tirne. Therefore, the estimates of the average cell cycle time for FGF-responsive

stem cells may hide the variability in ce11 cycle times that exist (Acklin and van der Kooy,

1993; Reznikov et al., 1997).

At E14, there are two populations of stem cells, an FGF-responsive stem ce11 and

a EGF-responsive stem cell. Approximately 40-80% of al1 stem cells are proliferating at

El4 with estirnated ce11 cycle times of 26 h n for FGF-responsive stem cells and 34 hrs

for EGF-responsive stem cells (0.4(Tc-T,) = 12 hours, Tc-T, = 30 hours, and T, = 4

hours). Between E 14 and E 17 the increase in the FGF-responsive stem ce11 population is

2-fold. The entire population of FGF-responsive stem cells would have to undergo only

one symmetric division to yield the number of FGF-responsive stem cells observed at

E 17. Given the estimates of ceIl cycle times at El4 and assuming the stem cells divide

throughout the three day period, it can be concluded that between El4 and El7 the

majority of divisions of FGF-responsive stem cells would be asymmetric. The EGF-

responsive stem ce11 population increases 5-fold between E 14 and E17, but its cell cycle

time remains relatively constant over this time (34-38 hrs). The population would have

to undergo between two and three symmetric divisions in 72 houn to generate the

numbers of EGF-responsive stem cells observed at E17. Given a 34 hrs ce11 cycle time

for EGF-responsive stem cells at E14, most of the EGF-responsive stem ce11 divisions

would need to be symmetrical to generate the 5-fold increase in stem cells between El4

, and E17. Altematively, if some EGF-responsive stem cells go through asymmetrk

divisions between El4 and El7 (generally only one EGF-sensitive stem ce11 and one

differenhating progenitor per division), then the EGF-responsive stem ce11 population

also could be expanded by way of the FGF-responsive stem cells generating EGF-

responsive stem cells through asymmetric divisions.

At E17, the FGF-responsive stem ceii population has a ce11 cycle time of

approximately 48.4 hours (0.27(Tc-T,) = 12 houn, Tc-T, = 44.4 hours, and Ts = 4 hours),

whereas the EGF-responsive stem ce11 population has a ceIl cycle time of approximately

38.3 hours (0.35(TC-T,) = 12 hours, Tc-T, = 34.3 hours, and Ts = 4 hours). It is not

known if either of these stem ceIl populations continue to increase in absolute numben

after E17, but it is apparent that they are lengthening the time of their ce11 cycles and

possibly becoming relatively quiescent, as they are in the subependyma of the adult

mouse forebrain (Morshead et al., 1994). Indeed, even when forcd to divide by depleting

their constitutively proliferating progeny, the adult stem cells still divide only

asymmetrically in vivo to renew themselves and to replenish the constitutively

proliferating population (Morshead et al., 1998). However, adult stem cells retain the

ability to divide symmetrically in the adult forebrain in vivo, as evidenced by the increase

in adult lateral ventricle subependymal stem cells afier direct infusion of EGF or FGF2 in

the lateral ventricle (Craig et al., 1996; Martens and van der Kooy, 1998).

DISCUSSION

In the present paper I determined the kinetics of neural stem cells by injecting

animals with hi@ doses of 'H-T~R to kill off proliferating cells and then culturing the

cells in the clonal neurosphere assay (where one sphere represents the progeny of a single

neural stem ceIl). According to my hypothesis if stem cells were proliferating at the time

of exposure to 'H-T~R, then they would die and faii to form neurospheres in vitro. Such

an approach may be the only way to study the proliferation kinetics of small

subpopulations of neural stem cells [<1% of germinal zone cells (Tropepe et al., 1999)J

within the larger population of embryonic germinal zone precurson. The results showed

that during embryonic forebrain development the length of the ceIl cycle and mode of

division of two separate populations of neural stem cells (EGF- and FGF-responsive)

change over time. The length of the cell cycle of FGF-responsive stem cells increases with

increasing embryonic age and their mode of division switches from king primarily

symmetric at El l to king primarily asymmetric by E14. EGF-responsive stem cells arise

behveen El 1 and E 14, have a similar relatively long ce11 cycle time at El4 and E 17, and

this EGF-responsive stem ce11 population increases in size by asymmetric divisions of

FGF-responsive stem cells and by symmetric divisions of EGF-responsive stem cells

themselves.

The ce11 cycle time of EGF-responsive stem cells seems to be relatively long

(estimated at over 30 hrs) soon after they appear at El4 and this longer ceIl cycle time

& 2 appears to be maintained by the EGF-responsive stem cells at E17. This contrasts with ,.-

the drarnatic increase in cell cycle time of the FGF-responsive stem cells (based on the in

vivo kill experiments). The very different cell cycle time behaviours of EGF and FGF-

responsive stem cells further support the data showing that the embryonic EGF- and

FGF-responsive stem cell populations are additive and separate populations of neural

stem cells (Tropepe et al., 1999).

Based on the depletions in the numben of neurospheres that are generated in

vitro, up to 88%, 60%, and 27% of al1 of the FGF-responsive stem cells are proliferating

within a 12 hour period at E 1 1, E 14 and El 7, respectively. These data suggest that the

length of the FGF-responsive neural stem ceIl cycle is inaeasing with increasing

embryonic age. This conclusion is supported by two further findings. First, the dose of

'H-T~R used to inject mice at each embryonic age was sufficient to kill off stem cells

proliferating in vivo. Second, more injections of 'H-T~R injected over a longer period of

time resulted in a further depletion in the number of neurospheres generated m vitro in

response to FGF2. At E 17 the incorporation of 3 ~ - ~ d ~ into the DNA of striatal germinal

zone cells and the depletion in FGF-generated neurospheres was signiticantly less than

that at El4 or E l 1, raising the possibility that the dose of 3 ~ - ~ d ~ employed was

insmcient at E17. However, doubling the dose of 'H-T~R at El7 did not fûrther increase

the depletion in the number of neurospheres that fomed nor further increase the

incorporation of the nucleotide in the DNA of proliferating cells. Therefore, at El7 it is

not that ' H - T ~ R is ineffective at killing labeled cells, but that the results reflect the fact

, that only a fraction of the cycling stem cells are in S-phase at the time of exposure to 'H- .*''

TdR.

Further support for the hypothesis that the cell cycle duration of the FGF-

responsive stem ce11 was increasing at El4 compared to at El 1 m e from cornparing the

effects of injections of 'H-T~R for 12 and 20 hours at E14. After 20 hours of injections

there was a greater depletion obsemed in the number of neurospheres generated in FGF2

than after 12 houn of injections. Assuming single population kinetics, I suggest that the

length of the cell cycle of FGF-responsive stem cells increases from approximately 17.2

houn at El 1 to approximately 26 houn at E14. However, these results also suggested

that there are actually 2 populations of proliferating FGF-responsive stem cells, one

majority population with a shorter ceIl cycle time (19.4 hours, 15.4 houn + 4 hours S

phase) and a second minority population with a longer ceIl cycle time (44 hours, 40 hours

+ 4 houts S phase). It is possible that the second population of FGF-responsive stem

cells represents a small fraction of the stem cells that have become relatively quiescent, as

stem cells exist in the adult forebrain subependyma (Monhead et al., 1998). Whether or

not there are two proliferating populations of FGF-responsive stem cells, my results

suggest that the average ce11 cycle time of the small subpopulation of neural stem cells

increases over embryonic time in a rnanner similar to the increase in the estimates of the

average ceIl cycle times of the overall population of germinal zone cells giving rise to the

striaturn (10- 12 hr fiom E 1 1 to E 12) (Bhide, 1996) and cortex (8- 1 8 hr fiom E 1 1 to E 16)

(Takahashi et al., 1995). Although this may imply that some geneml embryonic factor

, may be increasing ceIl cycle tirne throughout the genninal zone over embryonic .+

development, it is important to point out that the average estimates of cell cycle

parameten for the entire germinal zone population hide some large differences in ceIl

cycle times arnong different proliferating progenitor clones at both El4 (Reznikov et al.,

1997) and El 7 (Acklin and van der Kooy, 1993). Indeed, our estimates of the absolute

ce11 cycle times for both populations of neural stem cells at each embryonic age were

generally longer than the average estimates made for the overall populations of forebrain

germinal zone cells at similar ages. Furthemore, variability within and between individual

experirnents and evidence for a second population of stem cells having a really long cell

cycle time suggest that there is a certain amount of heterogeneity in ceIl cycle times within

the small proportion of germinal zone cells that are stem cells. Still, this does not affect

my conclusion that within the FGF-responsive stem cell population there is a signifiwt

lengthening of the ce11 cycle time with increasing embryonic age. Lengthening of the ce11

cycle time is thought to be a function of an increase in the duration of G 1 as the rest of

the ceIl cycle parameters remain relatively constant over time (Waechter and Jaensch,

1972; Caviness et al., 1995). An increase in the length of G1 provides an opportunity for

stem cells to respond to changes in their environment by, for instance, switching from

symmetric to asymmetric divisions.

At El4 there is a switch in the mode of division of the FGF-responsive stem cells

from pnmarily symmetric division to primarily asymmetric division. This switch in the

mode of division of FGF-responsive stem cells was based on the estimates of the ce11

?

cycle times and the estimates of the sizes of the population at each embiyonic age. The .id-

size of this stem ceIl population was detemined by estimating the total number of

neurospheres that would fom in vitro from the striata dissected from a saline treated

embyro at each age. As a result, like the estimates of ceIl cycle times, these estimates are

variable, and therefore can only be used to analyze the relative changes in the size of each

stem ce11 population over time. Between El 1 and El4 EGF-responsive stem cells are

generated. There is evidence that the EGF-responsive population is the progeny of FGF-

responsive stem cells (Tropepe et al., 1999). The switch to asymmetric division by FGF-

responsive neural stem cells may be a result of the demand for the production of EGF-

responsive stem cells. What signal(s) is responsible for such a switch in the mode of

division remains unclear.

In the developing forebrain, there is evidence t hat EGF-receptor immunostaining is

greatest in the subventricular zone, whereas FGF-receptor 1 mRNA levels are most

intense in the ventricular zone (Eagleson et al., 1996; Wanaka et al., 1991). Early in

embryogenesis only FGF-responsive stem cells reside in the ventricular zone. The

generation of EGF-responsive stem cells coincides with the formation of the

subventricular zone. Given that the germinal zone contains two distinct, lineage-related

stem cells and that the growth factor receptors appear to be differentially expressed

between the ventricular and subventricular zones, I hypothesize that FGF-responsive

stem cells seed the subventncular zone with EGF-responsive stem cells, and therefore

would reside in distinct proliferative zones. Indeed, preliminary evidence suggests that

more neurospheres are generated in response to EGF than to FGFZ fiom dissections of

El4 striatal subventricular zone tissue, whereas more neurospheres are generated in

response to FGFZ than to EGF h m dissections of El4 striatal ventncular zone tissue

(Tropepe and van der Kooy, unpublished observations). This would suggest that FGF-

responsive stem cells undergo asymmetric divisions in the ventricular zone, generating one

FGF-responsive stem ce11 and one EGF-responsive stem d l , and then the EGF-

responsive neural stem ce11 migrates through the ventncular zone to reside in the

subventricular zone.

Several factors are involved in determining if cells will undergo symmetric or

asymmetric divisions (for review see Jan and Jan, 1998). The asymmetric division of the

FGF-responsive stem ce11 may be regulated by m-Numb and its interactions with the

Notch 1 receptor (Guo et al., 1996; Zhong et al., 1996). Notch 1 has been found to be

asyrnmetrically localized in cortical progenitors. Cells undergoing asyrnmetric (cleavage

furrow is horizontal to wall of laterd ventricle) division were found to have the Notch 1

receptor segegated to the basilar daughter ce11 (Chenn and McConnell, 1999, whereas m-

Numb is segregated to the apical daughter ce11 (Zhong et al., 1996). It is possible that m-

Numb and Notch 1 may be regulating the modes of division of neural stem cells and the

fates of their progeny, although the prevalence of their expression suggests they must

have roles in the division of the more massive downstream population of germinal zone

progenitors. From the perspective of neural stem cells, 1 might suggest that during

asymmetric division of an FGF-responsive stem cell, m-Numb would be retained within

, the FGF-responsive stem cell, while Notch 1 would be inherited by the EGF-responsive -5'-

stem ceIl to help keep it in an undifferentiated state as it migrates through the ventricular

zone to the subventricular zone.

My data suggest that the small populations of striatal stem cells may change corn

symmebic to asymmetric divisions during embryonic development. In the adult

forebrain, neural stem cells divide asymmetrically, even after killing the constitutively

proliferating population (Monhead et al., 1998). However, they are still capable of

undergoing symmetric divisions in vitro (more than one new neurosphere is generated

afier passaging single neurospheres generated in either EGF or FGFZ) and in vivo

following the infusion of growth factors into the lateral and fourth ventricles of the adult

mouse brain (Craig et al., 1996; Martens and van der Kooy, 1998). This suggests that

cells undergo asymmetric division unless there is some positive factor within the

environment to tell them to divide syrnmetrically (Morshead et al., 1998).

Acklin SE, van der Kooy D (1993) Clona1 heterogeneity in the germinal zone of the

developing rat telencephalon. Development 1 18: 1 75- 192.

Altman J, Bayer SA (1995) Atlas of the prenatal rat brain development. Boca Raton, FL:

CRC Press.

Becker -45, McCulloch EA, Siminovitch L, Till JE (1965) The effect of diffenng demands

for blood cell production on DNA synthesis by hemopoietic colony-forming cells of

mice. Blood 26:296-3 12.

Bhide PG (1 996) Ce11 cycle kinetics in embryonic mouse corpus striatum. J Comp Neurol

374: 506-522.

Bj ursell G, Reichard P ( 1973) Effects of thymidine on deoxyribonucleoside triphosphate

pools and deoxyribonucleic acid synthesis in Chinese hamster ovary cells. J Bi01 Chem

248:3904-3909.

Caviness VS Jr., Takahashi T, Nowakowski RS (1995) Numbers, time and neocortical

neurogenesis: a geneml developmental and evolutionary rnodel. Trends in Neurosci

1 8:379-383.

, 5

Chen A, McCo~ell SK (1 995) Cleavage orientation and the asymmetric inhentance of

Notch 1 immunoreactivity in marnrndian neurogenesis. Ce11 82:63 1-64 1.

Craig CG, Tropepe V, Morshead CM, Reynolds BA, Weiss S, van der Kooy D (1996) ln

vivo growth factor expansion of endogenous subependymal neural precursor ce11

populations in the adult mouse brain. J Neurosci 16:2649-2658.

Drew RM, Painter RB (1959) Action of tritiated thymidine on the clonal growth of

mammalian cells. Rad Res 1 1535-544.

Eagleson KL, Fem RT, Levin P (1 996) Complementary distribution of collagen type IV

and the epidemal growth factor receptor in the rat embryonic telencephalon. Cerebral

Cortex 6: M W W

Frederiksen K, McKay RDG ( 1988) Proliferation and differentiation of rat

neuroepithelial precunor cells in vivo. J Neurosci 8: 1 144- 1 15 1.

Gritti A, Parati EA, Cova L, Frolichsthal P, Galli R, Wanke E, Faravelli L, Morassutti DJ,

Roisen F, Nickel DD, Vescovi AL (1996) Multipotential stem cells fiom the adult mouse

brain proliferate and self-renew in response to basic fibroblast growth factor. J Neurosci

16: 1091-1 100.

Guo M, Jan LY, Jan YN ( 1996) Control of daughter ce11 fates during asymmetric division;

interaction of Numb and Notch. Neuron 17:27-4 1.

Hall PA, Watt FM (1989) Stem cells: The generation and maintenance of cellular

diversi ty . Development 1 O6:6 19-633.

Hasthorpe S, Harris AW (1979) Growth control of hemopoietic colony-foming cells by

thymidine and deoxy cytidine. In Experimental Hematology Today . SJ Baurn and GD

Ledney, Eds. Springer-Verlag NY, pp. 1 1 - 18.

Jan YN, Jan LY (1998) Asymmetric cell division. Nature 392:775-778.

Lajtha LG, Pozzi LV, Schofield R, Fox M (1969) Kinetic properties of haemopoietic

stem cells. Ce11 Tissue Kinet 2:39-49.

Luskin MB, Pearlman AL, Sanes JR ( 1988) Ce11 Lineage in the cerebral cortex of the

mouse studied in vivo and in vitro with a recombinant retrovirus. Neuron 1 : 635-647.

Luskin MB, Parnavelas JG, Barfield JA (1993) Neurons? astrocytes, and

oligodendrocytes of the rat cerebral cortex originate fiom separate progenitor cells: An

ultrastructural analysis of clonally related cells. J Neurosci 13: 1730- 1750.

Luskin MB, McDermott K (1994) Divergent limages for oligodendrocytes and astrocytes

originating in the neonatal forebrain subventncular zone. Glia 1 1 : 2 1 1-226.

Martens DJ, van der Kooy D (1998) In vivo infusion of EGF, bFGF, and Heparin

increases proliferation of cells lining the foiirth ventricle and central canal of the adult

mouse brain. Soc Neurosci Abstr 24: 1528.

Morshead CM, van der Kooy D (1992) Postmitotic death is the fate of constitutively

proliferating cells in the subependymal layer of the adult mouse brain. J Neurosci 12:249-

256.

Morshead CM, Reynolds BA, Craig CG, McBumey MW, Staines WA, Momsatti D,

Weiss S, van der Kooy D (1994) Neural stem cells in the adult mamrnalian forebrain: A

relatively quiescent subpopulation of subependyrnal cells. Neuron 13: 1 O7 1 - 1082.

Morshead CM, Craig CG, van der Kooy D (1998) In vivo clonal analyses reveal the

properties of endogenous neural stem ce11 proliferation in the adult mamrnalian forebrain.

Development 2 25 : 225 1-226 1.

& ,'. Monhead CM, Tropepe V, Hadjantonakis A-K, Nagy A, van der Kooy D (1998) The . - . . -

non-cell autonomous sphere forming ability of neural stem cells. Soc Neurosci Abstr 24:

1528.

Nowakowski RS, Lewin SB, Miller MW ( 1989) Bromodeoxyuridine

immunohistochemical determination of the lengths of the ceil cycle and the DNA-

synthetic phase for an anatomically defined population. J Neurocyt 18:3 1 1-3 18.

Painter RB, Drew RM, Hughes WL (1958) Inhibition of HeLa growth by intranuclear

tritium. Science 127: 1244- 1245.

Potten CS, Loeffler M (1990) Stem 41s: Attributes, cycles, spirals, pitfalls, and

uncertainties. Lessons for and from the crypt. Development 1 10: 100 1 - 1020.

Reynolds BA, Weiss S ( 1992) Generation of neurons and astrocytes from isolated cells of

the adult mammalian central nervous system. Science 255: 1707- 1710.

Reynolds BA, Weiss S (1996) Clonal and population analyses demonstrate that an EGF-

responsive marnmalian embryonic CNS precunor is a stem cell. Dev Bi01 1 75: 1 - 13.

Reynolds BA, Tetzlaff W, Weiss S (1992) A multipotent EGF-responsive striatal

embryonic progenitor ce11 produces neurons and astrocytes. J Neurosci 12:4565-4574.

Reznikov K, van der Kooy D (1995) Variability and partial synchrony of the cell cycle in

the germinal zone of the early embryonic cortex. J Comp Neuroi 360536454,

Reznikov K, Acklin SE, van der Kooy D (1997) Clonal heterogeneity in the early

embryonic rodent cortical germinal zone and the separation of subventricular from

ventricdar zone lineages. Dev Dynamics 2 10: 328-343.

Skoff RP (1990) Gliogenesis in rat optic nerve: Astrocytes are generated in a single wave

before oligodendrocytes. Dev Biol 139: 149- 168.

Takahashi T, Nowakowski RS, Caviness VS Jr. (1993) Ce11 cycle parameten and patterns

of nuclear movement in the neocortical proliferative zone of the fetal mouse. J Neurosci

2 3: 820-833.

Takahashi T, Nowakowski RS, Caviness VS Jr. (1995) The ce11 cycle of the

pseudostratified ventncular epithelium of the embryonic murine cerebral wall. J Neurosci

1 5: 6046-605 7.

#

Tropepe V, Craig CG, Morshead CM, van der Kooy D (1997) Transforming growth $4

factor-a nul1 and senescent mice show decreased neural progenitor ce11 proliferation in the

forebrain subependyma J Neurosci 17: 7850-7859.

Tropepe V, Sibilia M, Cinina BO, Rossant J, Wagner EF, van der Kooy D (1999)

Distinct neural stem cells proliferate in response to EGF and FGF in the developing

mouse telencephalon. Dev Bi01 208: 166- 1 88.

Waechter RV, Iaensch B (1972) Generation times of the matrix cells during embryonic

brain development: An autoradiographic study in rats. Brain Res 46:235-250.

Wanaka A, Milbmdt J, Johnson, Ir EM (1 99 1 ) Expression of FGF receptor gene in rat

development. Development 455-468.

Zhong W, Feder JN, Ming-Ming J, Jan LY, Jan YN (1996) Asymmetric localization of a

mammalian numb homolog during mouse cortical neurogenesis. Neuron.17:43-53.