/ . Embryol. exp. Morph. Vol. 50, pp. 1-20, 1979
Printed in Great Britain © Company of Biologists Limited 1979
Production of erythropoietic
colony-forming units and erythrocytes during chick
embryo development: an attempt at
modelization of chick embryo erythropoiesis
By J. SAMARUT, 1 P. JURDIC 1 AND V. NIGON 1
From the Department of General and Applied Biology,
Claude-Bernard University Lyon I
SUMMARY
The enumeration of erythropoietic colony-forming cells in vitro has allowed us to complete
previous data on changes in the various erythroid cell populations during chick embryogenesis.
Erythrocytic colony-forming units in culture (CFU-cE) which are sensitive to avian
erythropoietin appear in the blastoderm as soon as the 24th hour of development. They
represent most likely precursors of the megalocytic erythropoiesis, and do not seem to
derive from stem cells common with normocytic erythropoiesis.
Data concerning vitelline normocytic erythropoiesis were analysed in a kinetic model
based on stochastic change of the stem cells. From this model it appears that 17-20 cell
divisions are required for differentiation of erythrocytes from stem cells.
INTRODUCTION
Vertebrate embryonic erythropoiesis occurs in several successive sites (review
in Nigon & Godet, 1976). In chick embryos, during the first 2 weeks, the yolk
sac is the main erythropoietic site. On the 12th day, erythropoiesis appears in
the marrow and develops mainly after the 16th day (Godet, 1974). Marrow
remains seemingly the unique erythropoietic organ after birth.
In the chicken, two kinds of erythrocyte precursors were identified. First, one
observes the CFU-M's (colony forming-unit in marrow), which give rise to
macroscopic colonies in marrow when injected into irradiated hosts (Samarut
& Nigon, 1975). Also seen are the CFU-cE's (colony forming unit in cultureerythropoietic), which develop erythrocytic clones in vitro when stimulated by
anaemic chick serum (Samarut & Nigon, 1976a). In vivo colonies derived from
CFU-M contain more than 104 cells while the colonies which grow in vitro
from CFU-cE have at most 150 cells.
1
Authors' address: Departement de Biologie generate et appliquee, Universite ClaudeBernard Lyon I, 43, boulevard du 11 novembre 1918, 69621 Villeurbanne, France.
2
J. SAMARUT, P. JURDIC AND V. NIGON
In past work, the kinetics of the various erythrocyte precursors have been
analysed especially in the mouse. However, in this animal erythropoiesis is
distributed among the various medullar sites, thus preventing a direct estimation
of the total medullar cell population. By contrast, in chicken embryos, the
erythropoietic activity is confined during more than half of the embryonic life
to the yolk sac, a continuous organ, in which the various populations can be
directly enumerated without any omission.
In previous work, we have determined the number of CFU-M during embryonic development of the chick (Samarut & Nigon, 19766; Jurdic, Samarut
& Nigon, 1978). The present work is devoted to the enumeration of CFU-cE in
the various embryonic hemopoietic sites. The results lead us to a coherent
kinetic model for the development of erythropoietic precursors.
MATERIAL AND METHODS
1. Preparation of cellular suspensions
(a) Bone marrow and yolk sac
Techniques for preparation of medullar and vitelline cells were identical to
those previously described (Samarut & Nigon, 1975, 1916b). Bone marrow was
mechanically dissociated and vitelline membrane was trypsinized. After several
washings, the cells were collected in TC 199 medium (Biomerieux) containing
50 % foetal calf serum (FCS, Gibco Biocult).
(b) Blastoderms
Most of the egg albumen was discarded through a hole in the shell. The
yellow yolk was gently poured into phosphate-buffered saline (PBS) pH 7-2 at
ambient temperature. The blastodermic area was cut off with fine scissors and
transferred into a Petri dish containing PBS. Under a dissecting microscope,
the vitelline membrane was detached from the blastoderm, which was then
agitated in the buffer to remove most of the attached yolk and finally transferred into a tube containing PBS with 10 % FCS. The blastoderms were then
mechanically dissociated by successive aspirations into a Pasteur pipette. The
cells were washed twice with PBS by spinning for 15min at 300 g. After the
last centrifugation, the pellet was harvested with TC 199 medium containing
50 % FCS. Since counting the cells in blastoderm suspensions was rather
difficult because of the presence of numerous yolk granules, concentration was
expressed as the number of blastoderms per ml of suspension, adjusted between
two blastoderms per ml for 3-somite embryos and 0-75 blastoderm per ml for
40-somite embryos.
2. In vitro culture
The plasma clot culture technique previously described (Samarut & Nigon,
1976 a) was slightly modified, especially by the use of chicken plasma instead
Erythrocytic colony-forming units in the chick embryo
3
of bovine plasma. Some batches of bovine plasma were found to be toxic for
chicken erythrocytes.
(a) Components of the culture medium
Foetal calf serum was decomplemented by heating at 56 °C for 30 min.
Among the various batches used, none was toxic in chick cell cultures.
TC 199 medium with Earle's salts (Biomerieux): this medium was buffered
at pH 7-2 with Hepes(JV-2-hydroxymethyl-piperazine-Af-2-ethanesulphonicacid)
at the final concentration of 20 mM.
/?-Mercaptoethanol was brought into the culture with the FCS to a final
concentration of 2 x 10~4 M.
Chicken plasma was collected by sterile technique from non-anaemic animals.
The plasma was extracted from blood collected on sodium citrate (one volume
of sodium citrate at 4 % (w/v) in PBS for nine volumes of blood). This plasma
was kept at 4° C and for no longer than 2 weeks. It was diluted with the same
volume of PBS just before use.
Anaemic chick serum was obtained from adult hens or roosters rendered
anaemic by bleeding according to the previously described method (Samarut &
Nigon, 1976«). This serum was dialysed against PBS and concentrated to half
the original volume, then stored at —20 °C.
Antibiotics: gentamicin was added to the cell suspensions to a concentration
of 10/tg/ml of final medium.
(b) Plasma clot culture
The different components of the culture medium were mixed in microtubes
in the following order: 20/d of FCS containing /?-mercaptoethanol 10~3 M,
20 /d of TC 199 medium, 30 /d of anaemic chick serum diluted with TC 199
medium, 20 /tl of cell suspension in TC 199 medium containing 50% FCS,
10 fi\ of twice diluted chicken plasma. The mixture was agitated on a Vortex
and two 40 /tl aliquots were layed into wells (7 mm diameter) of Linbro plates
IS-FB-96-TC. The plates were covered with adhesive tape Blenderm(3 M Co)
and incubated at 37 °C in humid air without CO2.
(c) Counting of the colonies
Cultures were observed on the 3rd day, to allow development of a maximum
number of colonies (Samarut & Nigon, 1976«). The erythrocytic colonies were
stained with benzidine. Into each well were poured three to four drops of a
mixture of 0-2 g benzidine and 0-4 ml hydrogen peroxide 110 volumes in 100 ml
ethanol 70 °C. The plates then remained in darkness for 20 min. The erythrocytic
colonies which exhibited a brownish colour were counted immediately afterwards. Colonies containing at least eight stained erythrocytes were scored.
J. SAMARUT, P. JURDIC AND V. NIGON
0
0
0-5
1-0
20
5-0
JO
20
20
0 0-5 10 20 50
10 20
Fig. 1. Dose response to anaemic chick serum for embryonic and adult CFU-cE.
Abscissa: anaemic chick concentration in percentage of the final volume. Ordinate:
erythrocytic colonies (or bursts) in percentage of the maximal number of induced
colonies (or bursts).
(A) Comparison between CFU-cE from yolk sac of 11-day-old embryo ( # ) and
CFU-cE from bone marrow of 2-month-old chicken ( • ) .
(B) Comparison between CFU-cE from 25- to 30-somite'4.blastoderm ( # ) and
CFU-cE from bone marrow of 2-month-old chicken ( • ) . The open circles (O)
refer to the corrected points for blastoderm assuming an endogenous production
of erythropoietin equivalent to 2 % of anaemic chick serum (see text for explanation).
(C) Comparison between CFU-cE ( # ) and BFU-E (A) from the marrow of
13-day-old embryo and CFU-cE from the marrow of 2-month-old chicken ( • ) .
The open circles (O) refer to the corrected values for erythrocytic colonies from the
embryonic marrow assuming an endogenous production of erythropoietin equivalent to 2 % of anaemic chick serum (see text for explanation).
(d) Sensitivity of the methods
Detection of CFU-cE. A fraction tS£ of the total number of the cells present in
one organ or one embryo were seeded into each culture well. The minimal
number of colonies which can be detected in each well is one colony. Consequently, the minimal number of CFU-cE detectable in the whole organ or
embryo was l/{§. For blastoderms harvested during the first hours of incubation: 00125 < tSJ, < 0-04. The method therefore did not allow the detection of
less than 25 to 80 CFU-cE's per embryo.
Detection of CFU-M. The number of CFU-M was estimated from the number
of colonies which developed in one tibia of grafted chickens previously irradiated. In the young embryo, the detection of CFU-M required the injection
into one host of all the cells harvested from several embryos (Jurdic et al. 1978).
Erythrocytic colony-forming units in the chick embryo
*
Fig. 2. Photomicrographs of erythrocytic colonies and burst. (A) Erythrocytic
colonies in a culture of embryonic marrow. (B) Erythrocytic burst in a culture of
embryonic marrow. Photographs were taken after staining of the culture with the
benzidine-H2O2 reaction.
Among the injected CFU-M's only a fraction/gives rise to medullar clones in
one tibia (Samarut & Nigon, 1975). Assuming that one colony is the minimal
number which can be detected in one tibia, the minimal number of detectable
CFU-M per embryo is 1/fx n, where n is the number of blastoderms injected
into one irradiated chicken. For young blastoderms 0023</<0-049 and
n < 8 (Jurdic et al. 1978). We therefore calculate that the method could not
reveal CFU-M numbers lower than three to five per embryo.
RESULTS
1. Sensitivity to erythropoietin of embryo and chicken CFU-cE's (Fig. 1)
The sensitivity to erythropoietin of CFU-cE's from various embryonic
haemopoietic tissues has been compared to that of medullar CFU-cE's from 1to 2-month-old chickens.
As the sensitivity curves can be slightly modified from one experiment to the
other by the use of different batches of the various components of the medium.
J. SAMARUT, P. JURDIC AND V. NIGON
12
24
36 48
1 1
1
1316
25
96
72
1
40
Hours
Somites
Fig. 3. Time course development of CFU-cE population in the blastoderm.
Abscissa: embryonic development in hours and in somitic stage. Ordinate: CFU-cE
per blastoderm.
each culture of embryonic CFU-cE was compared to a culture of adult marrow
CFU-cE grown in parallel using the same final medium.
Some embryonic tissues and especially the young blastoderms produce an
erythropoietic stimulating factor which seems identical to that found in the
serum of anaemic chicks (Samarut, 1978). The production of this factor by the
explanted tissues in plasma clot cultures is therefore revealed by the development of erythrocytic colonies even without supply of anaemic chick serum. The
dose-response curves obtained in such conditions have to be corrected taking
into account the concentration of the endogenous erythropoietic stimulating
factor.
In order to have comparative results between the various cultures, the numbers
of colonies which develop are plotted on the figures as percentages of the
maximum number of induced colonies.
In adult marrow cell cultures no colony develops without anaemic chick
serum. Colonies appear with at least 2 % anaemic chick serum and their number
is directly related to the logarithm of erythropoietin concentration. The maximal size of the colonies also increases with the concentration of stimulating
serum up to about 150 cells when the curve reaches a plateau. CFU-cE's from
11-day-old yolk sac (Fig. 1A) show exactly the same development except that
the observed points for vitelline cells are slightly above those corresponding to
marrow. Assuming that vitelline cells produce erythropoietin in vitro at a con-
Erythrocytic colony-forming units in the chick embryo
7
Table 1. Enumeration of CFU-cE in area opaca and area pellucida
CFU-cE per embryo
Development
stage
11-13 somites
16 somites
r
In area opaca
In area pellucida
23200
22800
25000
24000
0
0
1680
1840
centration equivalent to 0-5 % of anaemic chick serum, the curves for vitelline
cells and for medullar cells would be confounded.
In blastoderm and embryonic marrow cell cultures (Fig. 1B, C) erythrocytic
colonies containing at most 20 cells develop without supply of anaemjc chick
serum. For concentrations below 4 %, the number of colonies increases very
slightly. For higher concentrations, the evolution of the size and of the number
of colonies is similar to that of marrow-derived colonies. If we assume that
erythropoietin is produced in vitro at a concentration equivalent to 2 % of
anaemic chick serum, the points for blastoderm cells and for embryonic
marrow cells become very close to those observed for chicken marrow.
In embryonic marrow cultures another kind of erythrocytic colony develops
which represents aggregates of three or more colonies (Fig. 2). These clusters
are found even in cultures grown in larger wells such as 35 mm Petri dishes
where colonies are well dispersed so that they cannot be confused with fortuitous aggregates of single colonies. The cells which give rise to the colonies
and to the clusters respectively can be distinguished on the basis of their
membrane antigens (Samarut, Blanchet & Nigon, to be published), which
confirms that they represent distinct entities. These clusters require high levels
of anaemic chick serum to develop and their dose-response curve (Fig. 1C)
suggests a lower sensitivity to erythropoietin than the CFU-cE. These observations suggest similarity between these clusters and the '3-day burst' described
by Gregory (1976) in mouse haemopoietic cell cultures.
II. Development of CFU-cE population during embryonic development
CFU-cE were counted in cultures containing a saturating concentration of
anaemic chick serum determined from the curves of erythropoietin sensitivity.
(a) CFU-cE's in blastoderm (Fig. 3)
These cells were observed from the 3rd somite stage on (nearly 24 h of
incubation). No CFU-cE are found in unincubated blastodiscs or in 16 h or
20 h embryo which indicates that at 20 h the number of CFU-cE is less than
25 per embryo.
CFU-cE's harvested from less than 48 h embryos give rise in vitro to minute
colonies containing 20 cells or less. This CFU-cE population grows between
J. SAMARUT, P. JURDIC AND V. NIGON
Embryo
Fig. 4. Frequency of CFU-cE in yolk sac and in the marrow during embryonic
and postembryonic development. Abscissa: embryonic and post-embryonic
development stages in days (H = hatching). Ordinate: number of CFU-cE per 104
cells in yolk sac ( # ) and marrow ( • ) .
the 24th and the 32nd hour as an exponential with a 3-4 h doubling time. In
embryos of more than 25 somites, CFU-cE's develop colonies of 8-150 cells.
The corresponding CFU-cE population exhibits a doubling rate lower than
that of the previous population.
In order to state the origin of CFU-cE in the blastoderm we counted separately
the colonies produced by the cells from the area pellucida and the embryonic
anlage and those produced by the cells from the area opaca. The results are
presented in Table 1. Between the stages of 11 and 13 somites almost all the
CFU-cE's are found in the area opaca. In the 16-somite embryo the blood
stream may be responsible for some exchanges of CFU-cE between the area
opaca and the area pellucida.
b) CFU-cE's in the yolk sac and marrow (Figs. 4 and 5)
Vitelline cell preparations always contain some cells from the blood (Samarut
& Nigon, 19766). On the 6th day the frequency of the vitelline CFU-cE's has
been corrected taking into account the frequency of CFU-cE in the blood
(6 x 10~4). After the 6th day of incubation the frequency of CFU-cE in the
blood is too low to be measurable, therefore contamination of the vitelline
CFU-cE population by circulating CFU-cE was neglected.
The frequency of vitelline CFU-cE's falls sharply between the 6th and the
9th day of development, then remains nearly constant until the 13th day.
Afterwards it decreases to zero at hatching time (Fig. 4). From the 6th to the
Erythrocytic colony-forming units in the chick embryo
io 5 -
10
11
16
Embryo
Fig. 5. Development of total CFU-cE population in yolk sac and in marrow.
Abscissa: embryonic and post-embryonic development stages in days (H = hatching). Ordinate: total number of CFU-cE per yolk sac ( # ) and per tibia ( • ) .
13th day the total vitelline CFU-cE population increases as an exponential with
a 50 h doubling time. Then it decreases abruptly (Fig. 5).
In embryonic marrow the CFU-cE's appear with a high frequency on the
12th day and then decrease. Their frequency increases suddenly more than
10-fold during the 3 days before hatching, leading to a frequency of one
CFU-cE per 14 medullar cells. During the first week after hatching it decreases
back to values similar to those observed in embryonic marrow (Fig. 4). Total
growth of the embryonic marrow shows that the whole CFU-cE population
increases in successive waves with plateaus on the 16th day of incubation, and
during the end of the first week after hatching.
(c) Conclusion
The data at hand allow us to set up a synthetic representation of changes in
the various haemopoietic cell populations during development (Fig. 6). Values
concerning the CFU-M populations are taken from Samarut & Nigon. (19766)
and Jurdic et al. (1978). Data for the CFU-cE populations are from Figs. 3 and
5. Changes in normocytic and megalocytic cell populations, which include
mature erythrocytes and polychromatophilic erythroblasts, are taken from the
10
J. SAMARUT, P. JURDIC AND V. NIGON
0
2
4
6
8 10 12 14 16 18 20'
Embryo
H
2
4 6
Chick
8
Fig. 6. Development of erythropoietic cell populations in yolk sac and marrow.
A
A, CFU-cE population; #
it, CFU-M population; •
• j normocyte population; #
• , megalocyte population; © and <§), data calculated
from the model respectively for CFU-M population and for normocyte population.
Abscissa: embryonic and post-embryonic development stages in days (H = hatching). Ordinate: cell population number per yolk sac or per tibia.
results of Bruns & Ingram (1973) for the embryonic period and Blanchet
(personal commuaication) for the post-hatching period.
DISCUSSION
I. Development of megalocytic erythropoiesis
During the 2nd day of incubation, the graph of CFU-cE number appears in
direct continuity with the graph of megalocytes, suggesting that the CFU-cE's
Erythrocytic colony-forming units in the chick embryo
11
observed during this period correspond to megalocytic precursor cells. This
conclusion is in keeping with the results of Weintraub, Campbell & Holtzer
(1971) who observed that the first megaloblasts appear at the 25th hour of
incubation when the embryo contains about 2 x 104 haemocytoblasts, i.e. nearly
the same number as that of CFU-cE we found at the same stage.
Extrapolation of the observed data for CFU-M indicates a maximum of four
CFU-M at the 12th hour of incubation. It seems unlikely that such a small
number of stem cells could explain the occurrence of 4500 CFU-cE 12 h later:
this would require a generation time of about 1 h, which is highly unlikely.
These considerations lead to the hypothesis of no relationship between the first
CFU-cE's and the first CFU-M's appearing in the embryo.
Two types of hypotheses can be proposed to account for these results,
according to hypothesis of respectively monoclonal or polyclonal development
for megalocytic erythropoiesis.
According to the monoclonal theory, the whole megalocytic erythroblasts of
the embryo would arise from a unique haematopoietic stem cell. Weintraub et
al. (1971), using a mean generation time of 7-4 h estimated on the average
blastoderm cells, suggested that the determination of this stem cell occurs
nearly at the 7th post-fertilization mitosis, i.e. during the egg's transfer through
the oviduct. This monoclonal hypothesis predicts that the megalocytic line
develops as a synchronous cohort, which implies, for the CFU-cE, equivalence
between the cellular generation time and the doubling time of the cell population, i.e. nearly 3-5 h. With such a generation time one cell would need
approximately 42 h to produce the 3000 initial CFU-cE's observed at 24 h of
incubation. Such a monoclonal development would actually imply that the
initial determination takes place in the unincubated egg. This hypothesis should
suggest the existence of stem cells in non-incubated eggs. Demonstration of this
type of cell would require that they should be able to settle and grow when
transplanted in the marrow of an irradiated chicken.
Should the polyclonal theory be valid, one can imagine that the determination
occurs within the first 12 h of incubation. In this case it could be possible to
draw a model with an initial population of 25-200 determined cells and a
generation time of 3-5 h leading to the observed results. This hypothesis would
imply that determination of megalocytic primordial cells occurs directly at the
level of precursor situated between the CFU-M and the CFU-cE stages, to
explain the absence of CFU-M during the first hours of incubation.
II. Origin of normocytic erythropoiesis
It seems established that normoblastic erythropoiesis starts from a very low
number of CFU-M. These cells may arise consecutively to differentiating interactions as indicated by the observations of Miura & Wilt (1969). However, a
relationship between megalocytic and normocytic erythropoiesis could be
investigated at the level of common stem cells. Such an hypothesis would imply:
12
J. SAMARUT, P. JURDIC AND V. NIGON
(1) The existence of a stem cell stage in megalocytic erythropoiesis, which is
not presently proved.
(2) Differentiation parameters, inside this megalocytic stem cell compartment
such that most of the stem cells develop precursors of the megalocytic CFU-cE
and only a very small number into normocytic CFU-M.
Should such hypotheses not be valid, then one would have to conclude that
megalocytic and normocytic determinations occur independently in separate
precursors.
III. End of vitelline erythropoiesis
From the 10th day on, one observes breaks in the kinetic pattern of the
erythrocytic populations. These interruptions may have several explanations:
They might express a progressive exhaustion of the multiplicative abilities in
the vitelline erythropoietic cell population. In this case, development of medullar
erythropoiesis may imply that a distinct line becomes active. This distinction
between two erythrocytic lines was already suggested by the comparative
analysis of the kinetic development of vitelline and medullar CFU-M in
irradiated chickens (Samarut & Nigon, 1975, 1976&). The colonies which arise
from vitelline CFU-M's in the grafted irradiated chickens exhibit a kinetic
pattern of development different from that observed in colonies derived from
medullar CFU-M's.
They may express the exhaustion of the organ's ability to sustain erythrocytic
cell multiplication. In this case, one can imagine that the same stem cells
migrating from yolk sac to medullar sites will find there a more suitable
environment leading to the development of a medullar erythropoietic activity
from the 16th incubation day on (Godet, 1974). Under such an hypothesis, the
period between the 10th and the 16th day of incubation would correspond, for
erythropoiesis, to the slow degradation of the vitelline environment leading to
the vanishing of erythropoiesis near the 18th day, and to the multiplication and
progressive maturation from the stem cells colonizing the marrow.
IV. Attempt at an evaluation of the kinetic parameters of vitelline
normocytic erythropoiesis
We attempt to compile the whole data about normocytic erythropoiesis in a
unique model, by the use of the various under-mentioned hypotheses. Although
these hypotheses represent only a crude approach, the building of such a model
leads to numerical evaluations which can suggest more direct verification in the
future. Especially, the result of the calculation suggests the existence of ghost
precursor cells which cannot be directly revealed for lack of appropriate methods.
Erythrocytic colony-forming units in the chick embryo
CFU-M compartment
Number
of cell
divisions
0
CFU-cE compartment
T
T+U
13
E compartment
T+U+V
Fig. 7. Schematic representation of the erythropoietic differentiating pathway.
Unbroken arrows represent cell development concomitant with cell multiplication.
Dotted arrows represent cell maturation without mitosis. CFU-M, colony forming
unit in marrow; IC, intermediate cell; CFU-cE; colony forming unit in cultureerythrocytic; PE = proerythrocytic cell; MC = mother cell of erythrocytes; E =
erythrocyte. p, self renewal probability of CFU-M; q, differentiation probability of
CFU-M.
1. Building of the model
The building of the model is based on the following hypotheses, schematized
in Fig. 7.
(1) CFU-M develop according to a stochastic process in two directions: at
the end of each mitotic cycle either they progress into committed cells with a
probability q or they remain as CFU-M with a probability/? = 1 — q.
(2) CFU-M's which progress along the differentiating pathway are converted
into intermediate cells (IC).
(3) At the end of their Jth division, intermediate cells exhibit properties of
CFU-cE.
(4) At the end of the C/th division of the CFU-cE they become proerythrocytic
cells (PE) which have lost the ability to give rise to colonies in vitro.
(5) At the end of their Fth division, the proerythrocytic cells stop their
proliferation and become the mother cells of erythrocytes (MC) which enter
the compartment of E cells where they accumulate.
(6) It is assumed that all the cells exhibit the same generation time equal to g.
(7) It is supposed that, in the time lapse of the study, no destruction of
erythrocytes occurs. Actually, during embryonic life, loss of erythrocytes is
probably low with respect to their production.
Calculations based on these hypotheses will be found in the Appendix. It is
to be noted that according to these calculations, change of any cell compartment
shows two phases:
A first phase during which the compartment replenishes without discharging
into the following compartment.
A second phase during which each compartment is charged from the preceding compartment and discharges an outflow while its own cells are multiplying. The model shows that during this second phase changes in the various
2
EMB 50
14
J. SAMARUT, P. JURDIC AND V. NIGON
Table 2. Kinetic parameters estimated from the model
The values of p were determined from the curves on Fig. 8. The corresponding
values of g, T+ U and V were calculated respectively from equations (9) (10) and
(11) demonstrated in Appendix.
Duration of generations T+ U+V
in hours
Embryonic
period
2nd day
From the
3rd day on
Kinetic
parameter
96
g in hours
P
g in hours
P
T+U
V
3-5
0-68
4-9
0-59
130
4-3
108
3-7
0-69
5-7
0-61
140
5-2
compartments attain a dynamic equilibrium during which the population obeys
a linear relation when plotted in logarithmic scale. These linear functions allow
an evaluation of the numerical values of the parameters of the model.
2. Calculation from the experimental data
The graphs of CFU-M and CFU-cE population change exhibit linear
segments we can consider as parallel, in agreement with the proposed model.
These linear segments terminate by a slope change on the 10th day, which
suggests the appearance of a new kinetic pattern. In the erythrocyte graph, the
linear portion appears late, in agreement with the requirements of the model
(see Appendix) and is soon interrupted by the discontinuity on the 10th day.
The calculation of the kinetic parameters are described in the Appendix and
their values are presented in Table 2. The development of the erythropoietic
system between the 2nd and the 3rd day of incubation is characterized by the
increase of the cell generation time from 3-5-3-7 to 4-9-5-7 h and by a reduction
of the self renewal probability from 0-68-0-69 to 0-59-0-61. The calculations
show that 17-19 cell divisions are required for the differentiation of a CFU-M
into erythrocytes and that four to five divisions take place between the CFU-cE
and the mature erythrocyte compartments.
From the values of the parameters so determined and by assuming the
occurrence of four CFU-M at 24 h (p x M o ), the sizes of the various compartments can be calculated generation by generation. The results are plotted in
Fig. 6 simultaneously with the experimental data.
This model cannot be used for medullar erythropoiesis for the following
reasons. First, the kinetic curves for the various medullar cell populations do
not show parallel evolutions which could suggest that the displacement of the
erythropoietic activity between yolk sac and marrow is bound to adaptative
Erythrocytic colony-forming units in the chick embryo
15
changes in the values of several parameters. Second, medullar erythropoiesis is
distributed among various medullar sites, which prevents any exhaustive direct
enumeration of the precursor populations.
3. Discussion of the model and of the results
The building of this model is based on the classical principle of stochastic
development of haemopoietic stem cells currently proposed for the Mammals
(Till, McCulloch & Siminovitch, 1964; Vogel, Niewisch & Matioli, 1968, 1969;
Lajtha, Gilbert & Guzman, 1971). It must be noted that, in the present case,
we deal with a system whose components can be directly enumerated, without
the difficulties encountered in the mouse due to multiplicity of simultaneously
active erythropoietic sites. The calculations therefore require more limited
hypotheses. Our estimations were made assuming that no source of stem cells
other than the yolk sac and blood contributes to the development of vitelline
erythropoiesis during this period. However, the observations of DieterlenLievre (1975) and Dieterlen-Lievre, Martin & Beaupain (1977) tend to demonstrate the occurrence of haemopoietic cells in other embryonic sites. Further
experiments must be undertaken to determine if such intra-embryonic stem
cells effectively differentiate during the considered period.
In our model some compartments are defined by functional criteria, such as
the CFU-M and CFU-cE. A question arises from the limits between these
compartments:
Either the compartments overlap each other, which would imply that some
cells exhibit simultaneously CFU-M and CFU-cE properties; or the compartments are directly connected, i.e. the CFU-M keeps all its properties and loses
them at the moment it acquires the CFU-cE properties; or the compartments
are not contiguous and are separated by ghost compartments.
It is likely that, by the use of suitable hypotheses assuming different cell
generation times in the various compartments, one could calculate models
fitting to each of the three previous hypotheses. However, the occurrence of
parallel segments in the respective development of the CFU-M, CFU-cE and E
populations during the considered period seems to argue in favour of identical
(or not very different) cell generation times. The use of unequal cell generation
times would imply complex relationship between the kinetic parameters during
the various differentiation stages, so that our hypothesis appears as the most
simple one. According to this hypothesis, the occurrence of ghost compartment
appears therefore only as a logical requirement. Experimental proof of this
occurrence would require either the direct assessment of the cell generation time
or the elaboration of suitable methods for the detection of these ghost cells.
Presently, the occurrence of such compartments must be considered as a working
hypothesis.
The calculation shows that the more advanced CFU-cE's still undergo five
divisions before giving rise to erythrocytes. Should this number be maintained
16
J. SAMARUT, P. JURDIC AND V. NIGON
in vitro, these CFU-cE's would develop colonies containing at least 32 cells.
The less advanced CFU-cE's would give rise to more important colonies, and
it can be calculated that more than 5 % of the colonies should exhibit about 250
cells. Actually, the proliferation observed in vitro is much more limited. We
may estimate that, with respect to in vivo multiplication, it should be reduced by
at least two divisions. This phenomenon could explain why, in 2-day-old
blastoderm cultures, the megalocytic precursors, whose in vivo multiplicative
abilities are very low, develop minute colonies. On the contrary, in older
embryos, the colonies arise from less advanced CFU-cE and show a larger size.
Between the 2nd and 3rd day of embryogenesis, the development of the
erythropoietic system is characterized by a decrease of the value of p which
shows that the self renewal probability of the stem cell is not constant in each
tissue but can bejnodulated presumably by the physiologic environment.
APPENDIX
Theoretical calculations of the model
Let Mn, Cn and En be respectively the contents of CFU-M compartment,
CFU-cE compartment and Erythrocytes compartment at the wth cell generation.
It is assumed that at the beginning of generation zero the respective population sizes are Mo, Co = 0 and Eo = 0. The respective compartments are
expanding according to the hypotheses presented on page 13 and to the scheme
on Fig. 7.
The CFU-cE compartment starts to be supplied at the end of the Tth
generation and receives first a number of cells equal to qMQ2T. It is then filled
by multiplication of its own components and by the inflow from the preceding
compartment. From the generation T+ U on, the CFU-cE compartment starts
to deliver into the following compartment, supplying it, during the first generation, with a number of cells equal to qM02T+u.
The E compartment starts to be supplied from the generation T+ U+ V on,
and receives a first wave of cells equal to qM02T+u+v. At each following
generation the products of the preceding compartments accumulate in E compartment. It is admitted that there is no outflow from this last compartment
during the studied period.
It can be easily calculated that at generation n so that n>T+ U+V,
Mn=PM0(2p)\
(1)
The outflow rate Sn from the CFU-M compartment is
Sn = qM0(2p)\
The size En of population E is:
En = 2T+U+V
m=n-(T+U+V)
2
Sm
(2)
Erythrocytic colony-forming units in the chick embryo
17
which can be written:
En = Y^L Mo 2T+U+V \(2p)n+^T+u+^
- ll.
(3)
When n is rather high, it comes:
F
-
'
Mn
(4\
pT+u+v'
2p-\
The size Cn of CFU-cE population is:
Qi
=
2 [on_2i + 2 Sn_T-i + 2 on_y_2 + ... 2 ~ Sn-T-(u-i)]-
Which can be written:
Cn = MQ{2pY^rx
(5)
Cn = Mny£.
(6)
or
When U is rather great, this relation becomes:
n
C
Finally if Dn is the whole size of intermediate cell compartment and of
CFU-M compartment one calculates:
= M
lvl
n
p
pT+U
- C
~ ^
\ _pU
>'
w
In any preceding equation, we get the time function by writing n = (t-to)/g,
where t0 is the time of generation zero and g the duration of one generation.
From the experimental results we obtain:
ilog (2p) = log a.
(9)
Where log a is the slope of the curve representing the development of the
CFU-M population when plotted in logarithmic scale
C
-T2L = h =
Mn
Cn
(T+U+V)g
1
—
pT+u'
(2p-l)pv>
= d.
(12)
d = time lapse between the appearance of CFU-M and the appearance of
normocytes.
18
J. SAMARUT, P. JURDIC AND V. NIGON
-0-2
-
-0-4
-
0-7
0-8
Fig. 8. Evaluation of the probability p. The graphs of the functions y(p) and z(p)
were drawn according to the relations given in the Appendix. The functions z(p)
are calculated assuming b: 103 (calculated between days 4 and 8) and c — 50
(calculated on day 8). #
• , Function z(p) calculated during the 2nd incubation
day with a = 8-3"1; O
O, function z(p) calculated between days 2 and 9 of
embryonic development with a = 2-3 day 1 ; •
• , function y(p) calculated
with d = 96 h; •
*, functiony(p) calculated with d = 108 h. Abscissa: values
of probability p. Ordinate: values of y and z functions.
From this system we get
log 2
dlogp'd
log a
2(1-/7) '
1
lOg
bc(2p-\)
which can be solved graphically by drawing the functions:
log 2
1
dlogp
d
log a
(13)
(14)
%c(2p-\)
The actual value of p is given by writing y^v) = Z(p). The graphs of the functions
(13) and (14) are drawn on Fig. 8.
Erythrocytic colony-forming units in the chick embryo
19
Numerical calculations of the parameters
Strict application of the model predicts that the CFU-M compartment obeys,
from its origin on, a regular exponential law. However, this compartment shows
a higher slope at the beginning than later. Thus, we come to the conclusion that,
during the 2nd day of incubation, the parameters of the CFU-M compartment
must exhibit numerical values different from those acquired later. Considering
the exponential structure of the system, the disturbances so brought during a
limited period when the cell number is very low, have not justified the building
of a new model.
If we suppose that the CFU-M arise after 24 h of incubation, the appearance
of the normocytes 96-108 h later provides an evaluation of the minimal duration
needed for a number of generations equivalent to T+ U+V. The value of d
allows to calculate the functions y^ and Z(p). After this determination of/?, the
numbers of generations T+ U and V are assessed directly from equations (10)
and (11).
Assuming then that the duration and the number of divisions remain constant even during the 2nd day of incubation, the slope of the CFU-M curve
during this period allows to calculate a new value of p from the equation (9).
REFERENCES
G. A. P. & INGRAM, V. M. (1973). The erythroid cells and haemoglobins of the chick
embryo. Phil. Trans. R. Soc. B 266, 225-305.
DIETERLEN-LIEVRE, F. (1975). On the origin of haemopoietic stem cells in the avian embryo:
an experimental approach. / . Embryol. exp. Morph. 33, 607-619.
DIETERLEN-LIEVRE, F., MARTIN, C. & BEAUPAIN, D. (1977). Origin of erythropoietic and
lymphopoietic stem cells in the avian embryo, studied in interspecific chimaeras. Folia
biol., Pralia (P) 23, 373-376.
GODET, J. (1974). HbF synthesis in chicken embryonic and postnatal development. Studies
in various explanted erythropoietic tissues. Devi Biol. 40, 199-207.
GREGORY, C. J. (1976). Erythropoietin sensitivity as a differentiation marker in the hemopoietic system: studies of three erythropoietic colony responses in culture. / . Cell Physiol.
89, 289-302.
JURDIC, P., SAMARUT, J. & NIGON, V. (1978). Erythrocytic transplantable stem cells in young
chick blastoderms. Devi Biol. 64, 339-341.
LAJTHA, L. G., GILBERT, C. W. & GUZMAN, E. (1971). Kinetics of haemopoietic colony
growth. Br. J. Haemat. 20, 343-354.
MIURA, Y. & WILT, F. M. (1969). Tissue interaction and the formation of the first erythroblasts of the chick embryo. Devi Biol. 19, 201-211.
NIGON, V. & GODFT, J. (1976). Genetic and morphogenetic factors in hemoglobin synthesis
during higher Vertebrate development: an approach to cell differentiation mechanism.
Int. Rev. Cytol. 46, 79-176.
SAMARUT, J. (1978). Production of an erythropoietin-like factor by the young chick blastoderm. Devi Biol. (in press).
SAMARUT, J. & NIGON, V. (1975). Potentialites hematopoiietiques de tissus embryonnaires et
adultes analysees par greffe a des poussins irradies. /. Embryol. exp. Morph. 33, 259-278.
SAMARUT, J. & NIGON, V. (1976a). In vitro development of chicken erythropoietin sensitive
cells. Expl Cell Res. 100, 245-248.
BRUNS,
20
J. SAMARUT, P. J U R D I C AND V. N I G O N
J. & NIGON, V. (19766). Properties and development of erythropoietic stem cells
in the chick embryo. /. Embryol. exp. Morph. 36, 247-260.
TILL, J. E., MCCULLOCH, E. A. & SIMINOVITCH, L. (1964). A stochastic model of stem cell
proliferation based on the growth of spleen colony forming cells. Proc. natn. Acad. Set,
U.S.A. 51, 29-36.
VOGEL, H., NIEWISCH, H. & MATIOLI, G. (1968). The self renewal probability of hemopoietic
stem cells. /. CellPhysiol. 72, 221-228.
VOGEL, H., NIEWISCH, H. & MATIOLI, G. (1969). Stochastic development of stem cells.
/. theoret. Biol 22, 249-270.
WEINTRAUB, M., CAMPBELL, G. M. & HOLTZER, H. (1971). Primitive erythropoiesis in
early chick embryogenesis. I. Cell cycle kinetics and the control of cell division. /. Cell
Biol. 50, 652-668.
SAMARUT,
{Received 20 March 1978, revised 12 October 1978)