/ . Embryo/, exp. Morph. Vol. 46, pp. 135-146, 1978 135Printed in Great Britain © Company of Biologists Limited 1978

Modulation of alphafetoprotein synthesis in theearly postimplantation mouse embryo

By M. DZIADEK1

From the Department of Zoology, Oxford

SUMMARYThe visceral endoderm of mouse egg cylinders on the 7th and 8th days of gestation is

divided into the visceral embryonic (VE) endoderm cell population which synthesizesalphafetoprotein (AFP), and the visceral extra-embryonic (VEX) endoderm population whichdoes not synthesize AFP. Embryonic (E) and extra-embryonic (EX) ectoderm and visceralendoderm tissues were enzymically separated, reassociated in different combinations, andcultured in vitro for 48 h. The immunoperoxidase reaction on sections of cultured tissuesshowed that both VE and VEX endoderm cells synthesize high levels of AFP when culturedin isolation or in association with E ectoderm, but do not synthesize AFP when in closeassociation with EX ectoderm. Both 7th and 8th day VEX endoderm cells synthesize detect-able levels of AFP 12 h after isolation, and contain high levels by 24 h. It is concluded thatboth VE and VEX endoderm cells have the ability to synthesize AFP, but modulation ofexpression occurs through an inhibitory influence of the EX ectoderm.

INTRODUCTION

Initial differentiation of cells in the early mouse embryo appears to involvecell-cell or cell-environment interactions. After cavitation and expansion of theblastocyst, cells on the blastocoelic surface of the inner cell mass form a layer ofprimitive endoderm (Enders, 1971;Nadijcka & Hillman, 1974), apparently inresponse to an 'outside' position with respect to other cells (Gardner &Johnson, 1975; Rossant, 1975). The primitive endoderm contributes to thevisceral endoderm layer and to the parietal endoderm of Reichert's membrane(Snell & Stevens, 1966). The parietal and visceral derivatives differ in theirsynthetic properties. Parietal endoderm cells have been reported to synthesizea plasminogen activator (Strickland, Reich & Sherman, 1976), while visceralendoderm cells synthesize alphafetoprotein (AFP) (Wilson & Zimmerman;1976, Dziadek & Adamson, 1978). Studies on the localization and synthesis ofAFP during mouse embryogenesis showed that AFP synthesis is a propertyspecific to the visceral endoderm before liver formation (Dziadek & Adamson,1978). The immunoperoxidase reaction on tissue sections showed that AFP firstappears in some visceral embryonic (VE) endoderm cells of the 7th day embryo.On the 8th day AFP is confined to all YE endoderm cells, while the visceral

1 Author's address: Department of Zoology, South Parks Road, Oxford 0X1 3PS, U.K.

136 M. DZIADEK

extra-embryonic (VEX) endoderm cells do not react with the anti-AFP anti-serum. The visceral endoderm of 9th day embryos is also divided into AFP-positive and AFP-negative cell subpopulations, whereas all visceral endodermcells of the 10th day yolk sac contain AFP.

The clear-cut division of visceral endoderm cells into the AFP-positive andAFP-negative populations in 8th day embryos was very closely related to theirpositions over the embryonic (E) ectoderm and extra-embryonic (EX) ectodermtissues respectively. Such a correlation suggested that AFP synthesis in thevisceral endoderm may be dependant on an interaction with the underlyingtissue. The present study is an investigation of whether tissue interactions docontrol AFP synthesis in the visceral endoderm of the postimplantation mouseembryo. Control of AFP synthesis could be at two levels: at a primary level inwhich the tissue acquires a capacity for AFP synthesis (AFP synthetic pro-gramme); or at a secondary modulatory level in controlling the level of expressionof a programme which is already present. E and EX ectoderm tissues werereassociated with either VE or VEX endoderm to determine whether theE ectoderm exerts a positive influence over AFP synthesis, or whether EXectoderm exerts a negative influence, and at what level these influences operate.

MATERIALS AND METHODS

Isolation of embryonic tissues

Embryos used for all experiments were the F2 progeny of either C3H x 129Jor CBA x C57BL F\ crosses. The day of the vaginal plug was designated thefirst day of pregnancy. Uteri were dissected from mice on the 7th and 8th daysof pregnancy into the prewarmed phosphate-buffered saline (PBS, solution Aof Dulbecco & Vogt, 1954). Embryos were dissected from the decidual tissueinto Whitten's medium (Whitten, 1971) gassed with 5 % CO2, 5 % O2, 90 % N2

in a humidified incubator at 37 °C. Egg cylinders were isolated from the parietalendoderm and ectoplacental cone using tungsten needles. Two transverse cutswere made through the centre of the egg cylinders to produce completeembryonic and complete extra-embryonic fragments (see Dziadek and Adam-son, 1978). These were incubated in a solution of 2-5 % pancreatin and 0-5 %trypsin (Levak-Svajger, Svajger & Skreb, 1969) for 5 min at 4 °C. The layerswere separated by gentle pipetting using a glass pipette with bore diameterslightly smaller than the diameter of the tissue fragment. Isolated E and EXectoderm and VE and VEX endoderm tissues were washed and cultured ina-medium plus serum (see below) for £-1 h before reassociation to removeresidual enzyme activity.

Labelling of visceral endoderm by the uptake of melanin granules

Visceral endoderm cells can ingest melanin granules from surroundingmedium, and localization of melanin granules in sectioned tissue can be a usefulmarker for this cell type. Melanin granules were prepared from the eyes of male

Modulation of AFP synthesis 137CBA mice. Four whole eyes were dissected into approximately 2 ml ofWhitten's medium. The lenses were discarded and the heavily pigmented irisand retinal epithelia were shredded using fine forceps, thus releasing largequantities of melanin granules into the medium (method of R. L. Gardner,personal communication). This suspension was pipetted from the dish after thedebris had settled. Isolated VE and VEX endoderm layers were incubated inthe suspension of melanin granules for 3 h and then washed thoroughly inseveral changes of Whitten's medium. Melanin granules were freshly preparedfor each experiment.

Tissue culture

Isolated and reassociated tissues were cultured in microdrops of a-medium,lacking nucleosides and deoxynucleosides plus 10 % heat-inactivated foetal calfserum (Stanners, Eliceiri & Green, 1971), on bacteriological plastic Petri dishes(Sterilin Ltd., Richmond, Surrey), under liquid paraffin oil (Boots Pure DrugCompany, U. K.). Cultures were maintained in humidified 5 % CO2 in air, at37 °C. These culture conditions do not inhibit AFP synthesis in visceralendoderm cells (Dziadek & Adamson, 1978).

Immunoperoxidase reaction for AFP

The preparation of the anti-AFP antiserum and the tests for its specificityhave been described in a previous report (Dziadek & Adamson, 1978). AFPsynthesis by cultured tissues was determined by the immunoperoxidase reactionon tissue sections, using the Sainte-Marie technique for tissue fixation (Sainte-Marie, 1962) with Engelhardt's modification employing a mixture of 96 %ethanol with glacial acetic acid (99:1, v/v) for the fixative (Engelhardt, Goussev,Shipova & Abelev, 1971). Preparation of sections and the procedure forincubation in antisera and subsequent reaction with diaminobenzidine are asoutlined previously (Dziadek & Adamson, 1978). Control incubations weredone routinely with antiserum which had been absorbed with AFP. Very lowbackground levels of peroxidase activity were observed in these control sections,indicating negligible cross-reactivity of the antisera used, and the specificity ofthe anti-AFP antiserum for AFP.

RESULTS

AFP synthesis in visceral endoderm cells after reassociation with embryonic andextra-embryonic ectoderm

Cells of isolated tissues cultured in vitro can contain AFP only if they havesynthesized or are synthesizing AFP, since AFP is not present in the culturemedium for adsorption (Dziadek & Adamson, 1978). Isolated E and EXectoderm tissues do not contain AFP after culture, which indicates that thesetissues do not synthesize AFP and that a new layer of visceral endoderm cells

138 M. DZIADEK

MG

2A

MG

50 pim

1A

MG

MG

2C

MG

IB 2D

Modulation of AFP synthesis 139

does not reform from the isolated ectoderm (Dziadek & Adamson, 1978).All tissues known to synthesize AFP also accumulate AFP, and it is thereforeassumed in this study that absence of accumulation indicates that the tissueis not synthesizing AFP.

In in vitro studies on the control of AFP synthesis, it is necessary to havea reliable marker for visceral endoderm cells, which is independent of AFPsynthesis. Visceral endoderm cells can accumulate and retain melanin granulesin their cytoplasm, which are easily recognizable in histological sections.Preliminary studies showed that granules are taken up by 20-40 % of both VEand VEX endoderm cells after a 3 h incubation period. Between 5 and 15granules were initially present close to the cell membrane of each labelled cell(Fig. 1). Immunoperoxidase staining showed AFP to be present in VE endo-derm cells, but not VEX endoderm cells, immediately after incubation inmelanin granules. After 48 h in culture melanin granules were present through-out the cytoplasm of labelled cells, and both VE and VEX endoderm cellscontained high levels of AFP (Fig. 2). Melanin granules were not easily visualizedin cells with heavy peroxidase staining, but were present in adjacent controlsections. Uptake of melanin granules by visceral endoderm cells therefore doesnot inhibit their ability to synthesize AFP, and can be used as a marker for thesecells in studies on AFP synthesis.

The four tissue types (E and EX ectoderm, VE and VEX endoderm) wereisolated enzymically from 8 to 12 7th day embryos for each experiment. Visceralendoderm and ectoderm tissues were easily distinguishable under the transmitted-light microscope; the endoderm having a dark, granular appearance, and theectoderm appearing smoother and more translucent. Visceral endoderm tissuelayers also had the tendency to curl up after isolation. E and EX ectodermtissues were left intact, while the visceral endoderm layers were broken up intosmaller fragments by further pipetting. In three experiments visceral endodermcells were labelled with melanin granules, and in five experiments were leftunlabelled. In two initial experiments two ectoderm fragments were aggregatedwith two unlabelled visceral endoderm fragments, and in all subsequentexperiments single fragments of each tissue type were aggregated. In all cases

FIGURES 1 AND 2

Fig. 1. Immunoperoxidase reaction on sections of VE endoderm (A) and VEXendoderm (B) after culture in melanin granules (MG). Only VE endoderm cellscontain AFP, and melanin granules are present at the cell membrane of some cellsin both tissues.Fig. 2. Immunoperoxidase reaction on sections of VE and VEX endoderm tissueswhich were cultured for 48 h after incubation in melanin granules. Both VEendoderm (A) and VEX endoderm (C) contain high levels of AFP. Control sectionstreated with antiserum adsorbed with AFP show melanin granules (MG) present inboth VE (B) and VEX (D) endoderm cells. Calibration for B applies also for A;Calibration for D applies also for C.

140 M. DZIADEK

Table 1. The presence ( + ) or absence ( —) of AFP in visceral endoderm cells afterculture with embryonic or extra-embryonic ectoderm for 48 h, or in isolation for0 or 48 h

Experiments with melanin granules (MG) are indicated separately. The number ofobservations are summed from all experiments and are indicated in parentheses.

VEVE+MGVEXVEX+MG

Cultured inisolation,

Oh

+ (6)+ (8)- ( 6 )- (7)

Reassociatedwith

E ectoderm,48 h

+ (9)+ (5)+ (9)+ (6)

Reassociatedwith

EX ectoderm,48 h

-(13)- (U)*- (17)- (10)

Cultured inisolation,

48 h

+ (10)+ (8)+ (16)+ (8)

* In two aggregates (not included in total) AFP-positive VE endoderm cells wereattached as a clump to the EX ectoderm (see text).

the total mass of ectoderm was about three to four times that of the visceralendoderm. Ectoderm and visceral endoderm tissues do not reassociate readily,and aggregation was facilitated by frequently drawing the tissues togetherusing fine glass pipettes, or by initially limiting the size of the culture drop todraw the tissues together and incubating for a few hours before adding moremedium. In the majority of cases tissues had associated after 24 h in culture,and after 48 h the visceral endoderm was observed to have grown over theectoderm, but in most cases did not surround it completely. Very little tissuedegeneration was observed during the 48 h culture period, after which tissueaggregates were fixed for the immunoperoxidase reaction. Tn three experimentsrepresentative fragments of VE and VEX endoderm were fixed at zero time, todetermine the synthetic activity of these tissues before aggregation.

The results of these experiments are summarized in Table 1. Analysis of AFPsynthesis at the start of aggregation showed that VE endoderm cells containedAFP, whereas VEX endoderm cells were entirely unlabelled. When reassociatedwith E ectoderm, both VE and VEX endoderm cells contained high levels ofAFP after 48 h in culture in all experiments (Fig. 3). In aggregates with EXectoderm, no AFP-positive cells were present in the majority of cases. A fewouter cells in three aggregates were observed with very low levels of AFP. Inexperiments with labelled VE and VEX endoderm cells, the presence of melaningranules in outer cells of tissue aggregates indicated the presence of visceralendoderm cells (Fig. 4). In two cases VE endoderm cells had remained asa clump adhering to one side of the EX ectoderm, and many of these cells werenot in direct contact with EX ectoderm tissue. Immunoperoxidase stainingshowed that the outer endoderm cells contained AFP, presumably because theywere not in contact with EX ectoderm. In contrast, those visceral endodermcells which were labelled with melanin granules and which were in contact with

Modulation of AFP synthesis 141EX ectoderm, were AFP-negative (Fig. 5). In a few cases VE and VEX endodermcells had failed to associate with EX ectoderm and remained as free vesicles inthe culture drops. All of these visceral endoderm fragments were found tocontain high levels of AFP. Both VE and VEX endoderm tissues showed veryhigh levels of AFP after culture in isolation for 48 h in all cases.

The conclusion drawn from these results is that 7th day visceral endoderm fromboth regions of the egg cylinder has the inherent ability to synthesize AFP, butthat synthesis is inhibited by a contact-associated interaction with EX ectoderm.It appears that this interaction can inhibit AFP synthesis after it has alreadybeen initiated in the VE endoderm, and that contact must be continuous tomaintain inhibition in the visceral endoderm. The few cases where low levelsof AFP were still observed after association could still be due to later establish-ment of contact and delayed inhibition.

Initiation of AFP synthesis in isolated visceral endoderm

It is conceivable that the inhibition of AFP synthesis in visceral endodermcells in vivo becomes stabilized after a prolonged interaction with EX ectoderm,such that maintenance of the interaction is no longer necessary. To test whetherthis is the case, the capacity for AFP synthesis in the visceral endoderm of 8thday embryos was compared to that in 7th day embryos.

VE and VEX endoderm tissues were isolated enzymically from both 7thand 8th day embryos and cultured under identical conditions to compare theonset and progression of AFP synthesis. Embryos on the 7th day ranged frompre-primitive streak to early primitive streak stages of development, while8th day embryos were advanced egg cylinders with large embryonic cavities.Tissues were fixed for the immunoperoxidase reaction after 0, 6, 12, 24, 48 and72 h in culture. The results are summarized in Table 2.

AFP was present in VE endoderm cells from both 7th and 8th day embryosimmediately after isolation and at all stages during the culture period. AFP wasfirst observed in isolated VEX endoderm cells after 12 h in culture. The degreeof peroxidase staining was similar in VEX endoderm tissue from both 7th and8th day embryos. After 12 h low levels were present, which increased to veryhigh levels by 24 h. These results show that 8th day VEX endoderm does notdiffer from the 7th day VEX endoderm in capacity for AFP synthesis, and henceno stable inhibition of AFP synthesis has taken place over this time interval.However, unlike the VE endoderm in which most cells were AFP-positive after12 h in culture, patches of AFP-negative cells were often observed in the VEXendoderm cultures even after 72 h. This could be due to contamination by afew parietal endoderm cells which form a layer continuous with the VEX endo-derm in the intact embryo but do not synthesize AFP (Dziadek & Adamson,1978).

The results show that the interaction between EX ectoderm and visceralendoderm tissues does not induce a stable change in the capacity for AFP

10 EMB +6

142 M. DZIADEK

VE

VEX

50//m 50 //m

3A 3B

4A

EX

MG

VEX

50/im

4B

MG

50//m

Figs. 3-5. Immunoperoxidase reaction on sections of ectoderm and visceral endo-derm aggregates after 48 h in culture.

Fig. 3. VE (A) and VEX (B) endoderm cells associated with E ectoderm tissuecontain AFP.Fig. 4. VE (A) and VEX (B) endoderm cells associated with EX ectoderm tissue donot contain AFP, but some outer cells contain melanin granules.Fig. 5. An aggregate where VE endoderm cells remained as a clump attached to EXectoderm tissue. Cells closely associated with the EX ectoderm do not contain AFPbut are labelled with melanin granules, and cells in the adhering clump are AFP-positive. E, embryonic ectoderm; EX, extra-embryonic ectoderm; VE, visceralembryonic endoderm; VEX, visceral extra-embryonic endoderm; MG, melaningranules.

+ (4)+ (4)- (4)- ( 4 )

+ (9)+ (13)+ (7)+ (14)

+ (8)+ 02)+ (9)+ (12)

+ (15)+ (14)+ (18)+ (23)

+ (7)+ (14)+ (8)+ (10)

Modulation of AFP synthesis 143

Table 2. The presence ( + ) or absence ( —) of AFP in isolated visceral endoderm.tissue at various time intervals between 0 and 72 h in culture

The number of observations are indicated in parentheses.

Oh 6h 12 h 24 h 48 h 72 h

7th day, VE endoderm + (4)8th day, VE endoderm + (14)7th day, VEX endoderm - (4)8th day, VEX endoderm - (18)

synthesis by the visceral endoderm, but has a secondary modulatory role in theinhibition of phenotypic expression. Release from inhibition occurs whenvisceral endoderm is isolated from EX ectoderm.

DISCUSSION

In a study of the localization and synthesis of AFP during early mouseembryogenesis it was observed that the AFP-positive and AFP-negative cellsubpopulations of the visceral endoderm were closely associated with the typeof underlying ectoderm tissue (Dziadek & Adamson, 1978). The question arosewhether tissue interactions were involved in the synthetic activities of thevisceral endoderm. The present study establishes that this is the case. The resultsshow that VE and VEX endoderm cells from both 7th and 8th day egg cylinderscan synthesize AFP when growing in isolation from other embryonic tissues.Modulation of AFP synthesis occurs through an inhibitory interaction with EXectoderm tissue. The two subpopulations of the visceral endoderm therefore donot differ in their ability to synthesize AFP, but in the phenotypic expression ofthis ability. Initiation of AFP synthesis in VEX endoderm cells after releasefrom the influence of EX ectoderm tissue follows a similar pattern in cells fromboth 7th and 8th day embryos, which suggests that the mechanism for inhibitionof AFP synthesis may be the same over this developmental period.

A major criticism of studies involving tissue interactions in vitro concernsthe influence of culture conditions on phenotypic expression, as well as thepossible influence of enzymes used for tissue separation. However, in theinteraction studies reported here inhibition does not appear to be due to anydeficiencies of the culture conditions. In every experiment, visceral endodermcells were observed to synthesize AFP except when in direct contact with EXectoderm. Inhibition is not due to any factors in the culture medium which arereleased by EX ectoderm, since visceral endoderm cells growing in the sameculture drop but not associated with EX ectoderm were observed to containAFP. Very little cell degeneration was obvious during tissue culture, but even ifdegradation products were accumulating in the medium, these were not likelyto be causing the inhibition observed for the same reason as above. It is con-

144 M. DZIADEK

eluded that the absence of AFP in visceral endoderm cells which are associatedwith EX ectoderm tissue, both in vivo and in vitro, is due to an inhibitoryinfluence of the EX ectoderm.

The results presented in this report show that visceral endoderm cells canchange their phenotype in vitro when the tissue associations normally presentin vivo are altered. Whether such changes in phenotype occur in visceral endo-derm cells during the development of the intact embryo is not clear, since themovement of cells in the visceral endoderm layer with respect to underlyingtissues has not been studied. Observations on the relationship between pheno-typic expression and the type of tissue association in the in vitro experimentsreported here correspond to the observations on intact 7th and 8th day embryos(Dziadek & Adamson, 1978). However, in 9th day embryos the position ofAFP-positive and AFP-negative visceral endoderm cells does not correlateprecisely with the type of underlying tissue. Visceral endoderm cells which aresituated around the exocoelomic cavity now overlie a layer of extra-embryonicmesoderm cells. Only a small proportion of these visceral endoderm cells con-tain AFP, and these are positioned in a continuous band around the midgirth ofthe embryo (Dziadek & Adamson, 1978). From the results presentedinthis reportone would expect all these cells to be AFP-positive, since they are not in contactwith EX ectoderm tissue. It seems reasonable to propose that AFP-negativevisceral endoderm cells in this region have moved away from the influence ofthe EX ectoderm but have not yet initiated AFP synthesis. Migration of extra-embryonic mesoderm into the exocoelomic cavity and subsequent expansionof this cavity could create a change in tissue associations. From the observationsof initiation of AFP synthesis in vitro, cells which have moved from the influenceof EX ectoderm should synthesize low levels of AFP after 12 h. On the 10th dayof development all visceral endoderm cells of the yolk sac contain high levelsof AFP (Dziadek & Adamson, 1978), which substantiates such a proposal.However, this question needs further investigation.

The tissue interaction reported in this study occurs at the stage in mousedevelopment when only two germ layers are present, and is the first example ofmodulation of phenotypic expression in the early mouse embryo through aninhibitory influence. However, an earlier tissue interaction is thought to occurbetween the inner cell mass and trophectoderm of blastocysts (Gardner, 1971,1972; Ansell & Snow, 1974; Gardner & Johnson, 1975). The different influencesof the E and EX ectoderm is not surprising, since they have different origins inthe early embryo (Gardner & Johnson, 1975; Gardner & Papaioannou, 1975),and show different properties when grown in vitro and in ectopic sites (Diwan &Stevens 1975; Rossant& Ofer, 1977). It remains to be determined which propertiesof the EX ectoderm are involved in the inhibition of AFP synthesis in the over-lying visceral endoderm.

The mechanisms which control AFP synthesis in the visceral endoderm ofthe early embryo may be similar to controls of AFP synthesis in other tissues.

Modulation of AFP synthesis 145The second major site of AFP synthesis during embryogenesis is the foetal liver(Engelhardt et al. 1971; Abelev, 1971; Wilson & Zimmerman, 1976). Thedevelopmental mechanisms by which transitory foetal antigens such as AFPbecome inhibited after birth are not yet understood. Evidence that inhibition isnot completely irreversible comes from studies showing the reappearance ofAFP synthesis in mice with hepatomas (Abelev, 1971; Engelhardt et al. 1971)and during liver regeneration (Engelhardt et al. 1976). Studies on the controlsof AFP synthesis during embryogenesis and the maintenance of these controlsduring foetal and adult life may provide a basis for studies on the regulation ofgene expression in carcinogenesis.

I wish to thank Dr Chris Graham, Dr Eileen Adamson and Ms Jolanta Opas for criticalreading of the manuscript. I am supported by a Flinders University of South AustraliaOverseas Scholarship.

REFERENCES

ABELEV, G. I. (1971). Alphafetoprotein in ontogenesis and its association with malignanttumors. Adv. Cancer Res. 14, 295-358.

ANSELL, J. D. & SNOW, M. H. L. (1974). The development of trophoblast in vitro from blasto-cysts containing varying amounts of inner cell mass. / . Embryol. exp. Morph. 33, 177-185.

DIWAN, S. & STEVENS, L. C. (1975). Development of teratomas from the ectoderm of mouseegg cylinders. Proc. Am. Ass. Cancer Res. 16, 98-105.

DULBECCO, R. & VOGT, M. (1954). Plaque formation and isolation of pure lines with polio-myelitis viruses. / . exp. Med. 99, 167-182.

DZIADEK, M. & ADAMSON, E. (1978). Localization and synthesis of alphafetoprotein inpostimplantation mouse embryos. / . Embryol. exp. Morph. 43, 289-313.

ENDERS, A. C. (1971). The fine structure of the blastocyst. In Biology of the Laboratory Mouse,2nd ed. (ed. E. L. Green), pp. 205-245. New York: McGraw-Hill.

ENGELHARDT, N. V., GOUSSEV, A. I., SHIPOVA, L. J. & ABELEV, G. I. (1971). Immuno-fluorescent study of a-fetoprotein in liver and liver tumors. I. Technique of a-fetoproteinlocalization in tissue sections. Int. J. Cancer 7, 198-206.

ENGELHARDT, N. V., LAZAREVA, M. N., ABELEV, G. I., URYVAENA, I. V., FACTOR, V. M. &BRODSKY, V. Y. (1976). Detection of a-fetoprotein in mouse liver differentiated hepatocytesbefore their progression through S-phase. Nature, Lond. 263, 146-148.

GARDNER, R. L. (1971). Manipulations on the blastocyst. Advances in Biosciences 6, 279-296.GARDNER, R. L. (1972). An investigation of inner cell mass and trophoblast tissues following

their isolation from the mouse blastocyst. / . Embryol. exp. Morph. 28, 279-312.GARDNER, R. L. & JOHNSON, M. H. (1975). Investigation of cellular interaction and deploy-

ment in the early mammalian embryo using interspecific chimaeras between the rat andmouse. In Cell Patterning (Ciba Found. Symp. 29), pp. 183-196. Amsterdam: AssociatedScientific Publishers.

GARDNER, R. L. & PAPAIOANNOU, V. E. (1975). Differentiation in the trophectoderm andinner cell mass. In The Early Development of Mammals (2nd Symposium of the BritishSociety for Developmental Biology) (ed. M. Balls & A. E. Wild), pp. 107-132. CambridgeUniversity Press.

LEVAK-SVAJGER, B., SVAJGER, A. & SKREB, N. (1969). Separation of germ layers in presomiterat embryos. Experientia 25, 1311-1312.

NADUCKA, M. & HILLMAN, N. (1974). Ultrastructural studies of the mouse blastocyst sub-stages. /. Embryol. exp. Morph. 32, 675-695.

ROSSANT, J. (1975). Investigation of the determinative state of the mouse inner cell mass.II. The fate of isolated inner cell masses transferred to the oviduct. / . Embryol. exp. Morph.33, 991-1002.

146 M. DZIADEK

ROSSANT, J. & OFER, L. (1977). Properties of extra-embryonic ectoderm isolated from post-implantation mouse embryos. /. Embryol. exp. Morph. 39, 183-194.

SAINTE-MARIE, G. (1962). A paraffin embedding technique for studies employing immuno-fluorescence. / . Histochem. Cytochem. 10, 250-256.

SNELL, G. D. & STEVENS, L. C. (1966). Early embryology. In Biology of the Laboratory Mouse,2nd ed. (ed. E. L. Green), pp. 205-245. New York: McGraw-Hill.

STANNERS, C. P., ELICEIRI, G. L. & GREEN, H. (1971). Two types of ribosome in mouse-hamster hybrid cells. Nature, Lond. 230, 52-54.

STRICKLAND, S., REICH, E. & SHERMAN, M. I. (1976). Plasminogen activator in early embryo-genesis. Cell 9, 213-240.

WHITTEN, W. K. (1971). Nutrient requirements for the culture of preimplantation embryosin vitro. Advances in Biosciences 6, 129-139.

WILSON, J. R. & ZIMMERMAN, E. F. (1976). Yolk sac: site of developmental microhetero-geneity of mouse a-fetoprotein. Devi Biol. 54, 187-200.

(Received 27 January 1978, revised 12 April 1978)