Cell Differentiation in Isolated Inner Cell Masses of Mouse Blastocysts in Vitro: Onset of Specific Gene Expression

/. Embryol. exp. Morph. Vol. 53, pp. 367-379, 1979 357 Printed in Great Britain © Company of Biologists Limited 1979

Cell differentiation in isolated inner cell masses of mouse blastocysts in vitro: onset of specific gene expression

By MARIE DZIADEK1 From the Department of Zoology, University of Oxford, U.K.

SUMMARY Inner cell masses (ICMs) were isolated by immunosurgery from giant blastocysts formed by the aggregation of three morulae. A layer of endoderm cells formed on the outer surface of these primary ICMs in vitro. When this layer was removed by immunosurgery, a secondary endoderm layer formed. Alphafetoprotein (AFP) was used as a biochemical marker to characterize visceral endoderm formation in these cultured ICMs. The immunoperoxidase reaction on sections of ICMs cultured for intervals up to 120 h in vitro showed that some primary endoderm cells contained AFP, but these were always in the minority. The secondary endoderm layer, on the other hand, was composed of predominantly AFP-positive cells. It is concluded that the primary endoderm contains mainly parietal endoderm cells, while the secondary layer contains visceral endoderm cells. A model is proposed for the consecutive differentiation of parietal and visceral endoderm cell types from the ICM of mouse blastocysts.

INTRODUCTION The mouse blastocyst on the fourth day of gestation consists of two cell types: an outer layer of trophectoderm cells and an inner cell mass (ICM). On the fifth day of gestation a morphologically distinct epithelial-like layer forms on the blastocoelic surface of the ICM. This layer of cells consistitutes what has been termed the primitive enoderm (Enders, 1971; Nadijcka & Hillman, 1974) and is thought to be the stem cell population which gives rise to the parietal and visceral endoderm cells in the later egg cylinder. When ICMs are isolated from the trophectoderm layer by either microsurgery (Rossant, 1975) or immunosurgery (Solter & Knowles, 1975; Strickland, Reich & Sherman, 1976; Pedersen, Spindle & Wiley, 1977), a complete outer layer of endoderm forms within 24 to 48 h of culture. This endoderm layer appears to be composed of a mixture of parietal and visceral endoderm cells, using morphological criteria (Solter & Knowles, 1975; Pedersen et al, 1977). Pedersen et al. (1977) have shown that a second layer of endoderm can 1 Author''s present address: Division of Biology, Kansas State University, Manhattan, Kansas 66506, U.S.A.

24-2 368 MARIE DZIADEK 'regenerate' when the first is removed from giant ICMs formed from aggrega- tions of between 4 and 12 morulae. This second layer was morphologically similar to the first. To characterize the type of endoderm cells formed on isolated ICMs it is necessary to have tissue-specific markers for both parietal and visceral endoderm cells. Plasminogen activator has been used as a marker for parietal endoderm cells in differentiating ICM cultures (Strickland, et al., 1976), but recent studies have shown that plasminogen activator is also secreted by ectoderm, mesoderm and visceral endoderm tissues from the mouse embryo and is, therefore, unsuit- able as a tissue-specific marker (Bode & Dziadek, 1979). Visceral endoderm cells, on the other hand, synthesize alphafetoprotein (AFP), a fetal serum protein, which can be used as a specific biochemical marker to characterize this cell type in vivo and in vitro in the absence of fetal hepatocytes which are the only other source of AFP in later embryos (Dziadek & Adamson, 1978). Parietal endoderm cells do not synthesize or accumulate AFP at any stage of development (Dziadek & Adamson, 1978). Visceral endoderm cells only fail to synthesize AFP when they are closely associated with extra-embryonic ectoderm tissue (Dziadek, 1978) and perhaps all trophectoderm-derived tissues, and under such situations visceral endoderm cells would not be recognized. ICMs can be isolated free of all trophectoderm cells from fully expanded blastocysts by immunosurgery (Solter & Knowles, 1975; Handyside & Barton, 1977; Hogan & Tilly, 1978 b). Hence AFP can be used as a reliable biochemical marker to characterize visceral endoderm formation in cultured ICMs. In the present study a comparison is made of AFP expression in the first endoderm layer to be formed on isolated ICMs and the second which 'regene- rates' when the first is removed.

MATERIALS AND METHODS Giant blastocyst formation

Embryos used in all experiments were the F2 progeny of Fx (CBA x C57BL) natural matings. Giant blastocysts were formed by the aggregation of three 8-cell-stage embryos to increase the size of the ICMs for easy experimental manipulation and immunohistochemistry. 8-cell-stage embryos were recovered from pregnant females on the morning of the third day of gestation (9.00-12.00 h, the day of the copulation plug being designated the first day of pregnancy). Oviducts with the upper third of the uterus were dissected into pre-warmed and pre-equilibrated Whitten's medium (Whitten, 1971). Embryos were most often located in the last loop of the oviduct before the oviduct-uterine junction, and were released by tearing the loop open with fine watchmaker's forceps under a Wild M5 dissecting microscope. In some mice, embryos had already entered the uterus, and so in each case the upper part of the uterus was also torn open. Embryos were transferred to fresh Cell differentiation in isolated ICMs 369 Whitten's medium. The zonae pellucidae were removed by placing embryos in pronase at 4 °C, and incubating for 10-15 min at 37 °C (0-5 % pronase dialyzed against phosphate buffered saline, PBS, Solution A of Dulbecco & Vogt, 1954) followed by gentle pipetting in Whitten's medium. Morulae were aggregated in groups of three in microdrops (2-5 fi\) of Whitten's medium under paraffin oil (Boots Pure Drug Co., U.K., specially selected to be non-toxic) on bacteriological grade plastic petri dishes (Sterilin, Richmond, Surrey, U.K.) and cultured at 37 °C in a sealed box which had been gassed with humidified 5 % CO2, 5 % O2, 90 % N2. Cultures were maintained for 48 h until fully expanded blastocysts were formed. This was found to be approximately 12 h later than when single morulae were cultured in the same way.

Isolation of ICMs by immunosurgery The trophectoderm layer of blastocysts can be selectively killed using a two- step cytoxicity procedure which involves pre-incubation with anti-mouse antiserum followed by exposure to complement (Solter & Knowles, 1975). The trophectoderm cells are connected by tight junctions which prevent antibody binding to the inner cells, and so the ICM is protected from complement- mediated lysis.

(a) Preparation of anti-mouse antiserum

Conceptuses together with placental tissue were dissected from C3H and F2 (CBA x C57BL) female mice on the 13th—15th days of gestation. All tissue was washed, minced and homogenized in PBS. After low-speed centrifugation for 5 min, the supernatant was collected and used as the material for injection. A female rabbit was injected intradermally at monthly intervals, the first injection with the embryonic extract plus Freund's complete adjuvant (50:50 v/v) and the two subsequent injections with the embryonic extract plus Freund's incomplete adjuvant (50:50 v/v). Blood was collected one week after the final injection. The serum was heated at 56 °C for 30 min to inactivate rabbit complement and was left unabsorbed since no specificity was required.

(b) Immunosurgery procedure The anti-mouse antiserum was used at 1/10 dilution in Whitten's medium. Embryos were exposed to antiserum in a 0-5 ml volume for 30 min in a gassed incubator at 37 °C and subsequently washed three times in fresh Whitten's medium. Embryos were then exposed to 1/10 dilution of guinea-pig complement (Flow Labs, Ayrshire, Scotland) for 15 min and washed again in fresh Whitten's medium. ICMs were observed to round up after complement treatment. The lysed trophoblast layer was easily removed by pipetting the blastocysts through a fine-bore glass pipette, leaving a clean, smooth surface on the isolated ICMs. To test whether all trophectoderm cells had been removed, seven isolated ICMs were cultured on tissue-culture-grade petri dishes and observed over a 5-day

24-3 370 MARIE DZIADEK period for the outgrowth of trophoblast giant cells (Hogan & Tilly, 1978 a, b). No giant cells formed in these cultures, and it was therefore assumed that all trophectoderm cells were removed by the immunosurgery procedure. Culture oflCMs ICMs were cultured in microdrops (approximtely 5 jul) of a-medium, lacking nucleosides and deoxynucleosides, supplemented with 10% fetal calf serum (Stanners, Eliceiri & Green, 1971) under paraffin oil (Boots), gassed with humidified 5 % CO2 in air at 37 °C. Bacteriological grade plastic Petri dishes (Sterilin) were used to prevent adherence of ICMs, and hence maintain them in suspension culture. Cultures were maintained for up to five days, with medium being changed every two days. Analyses of AFP activity were made on each day of culture. In some cases the primitive endoderm layer was removed from ICMs by immunosurgery after 24-48 h in culture (Fig. 1). The procedure was identical to that described above for removing the trophectoderm layer.

Alphafetoprotein analyses ICMs were analyzed for cellular localization of AFP at 24 h intervals by the immunoperoxidase technique on tissue sections (Dziadek & Adamson, 1978). ICMs were fixed by the Sainte-Marie technique (Sainte-Marie, 1962) using Engelhardt's modification (Engelhardt, Goussev, Shipova & Abelev, 1971). After fixation ICMs were stained in 1 % Eosin Y (Sigma) for 30 sec and then dehydrated. This allowed visualization of the ICMs during the embedding procedure and identification of tissue within wax sections. The preparation of the anti-AFP antiserum and the tests for its specificity have been described previously (Dziadek & Adamson, 1978). The histological preparation of tissue sections and the procedure for incubation in antisera and subsequent reaction with diaminobenzidine are also outlined in the report cited above. Control incubations using AFP-absorbed antiserum were not done, since all previous controls on embryonic tissue had proven negative (Dziadek & Adamson, 1978; Dziadek, 1978) and it was important to detect all cells which contained AFP.

RESULTS Morphology of isolated ICMs 8-cell-stage embryos aggregated readily in approximately 90 % of cases and formed a large compacted morula within 6 h. The first signs of formation of a bJastocoelic cavity were observed 22-30 h after morula aggregation and fully expanded blastocysts developed within another 24 h (Fig. 1). In the cases when aggregation was not successful, two or three smaller Wastocysts were formed, often in a closely adhering group. These were not used for ICM isolation. ICMs from giant blastocysts were isolated routinely 48 h after morula aggregation (Fig. 1). Blastocysts at this time differed slightly in their degree Cell differentiation in isolated ICMs 371

Equivalent gestation day 3rd 4th 5th 6th 7th 9th 10th Morula Giant aggregation blastocyst

i Immunosurgery 1

i y Primary ICM (?) .. ^R\ Time in culture (h) 0 24 48 72 96 120

Immunosurgery 2

Secondary ICM Q » (Q) Time in culture (h) 0 24 48 72 96 Fig. 1. Experimental design for the formation of primary and secondary ICMs. of development. In all five blastocysts which were fixed and sectioned, a layer of endoderm-1 ike cells was present over the ICM, but in only two were cells observed adjacent to trophectoderm cells around the blastocoelic cavity, which may have been parietal endoderm cells. ICMs isolated from blastocysts will be called primary ICMs. These were observed to form a complete outer layer of endoderm cells in culture within 24 h after isolation in all cases, which was more distinct at 48 h. This layer will be called primary endoderm. After 48 h in culture, a cavity was observed to form in the inner core of the ICMs in over 75 % of cases. Development thereafter was fairly heterogeneous. Many ICMs remained as what appeared to be simple two- layered structures, while in some the internal cavity became very large, and development of several layers within the inner core was observed. The endoderm layer from primary ICMs was removed by immunosurgery after 24 h in most cases (Fig. 1) but in some after between 36 and 40 h, or after 48 h. Clean inner cores were isolated in all cases, which had a smooth outer surface similar in appearance to primary ICMs after their isolation. Such isolated ICM cores will be called secondary ICMs. Secondary ICMs isolated after 24 h or between 36 and 40 h after primary ICM culture regenerated a second layer of endoderm (secondary endoderm) in all cases. However, when immunosurgery was done at 48 h none of the five secondary ICMs was observed to regenerate a new layer but retained a fairly homogeneous appearance over the subsequent 3-day culture period. Secondary endoderm appeared as a thick, highly vesiculated layer within 24 h of secondary ICM culture. This secondary endoderm was removed by immunosurgery from eight ICMs and in no case was regeneration of a third endoderm layer observed by morphological criteria. 372 MARIE DZIADEK

Table 1. Number of primary and secondary ICMs showing different proportions of AFP-positive cells in the outer endoderm layer after different times in culture

Total Number of ICMs within each percentage number group of AFP-positive cells HTimp in A j. 11 lit- in v/o1f ICMs culture ICMs 0% 5-25% 25-50% 50-75 % 75-95 % 100% Primary Oh 4 4 — — — — Primary 24 h 9 9 — — — — — Primary 48 h 7 5* 2 — — — — Primary 72 h 6 — 6 — — — — Primary 96 h 4 — 3 1 — — — Primary 120 h 3 — — 3 — — — Secondary Oh 5 5 — — — — — Secondary 24 h 6 1 5 — — — — Secondary 48 h 7 — — 1 4 2 — Secondary 72 h 8 — — — 1 4 Secondary 96 h 5 — — — 1 2 2t * In two of these cases a few AFP-positive cells were observed in a layer underneath the outer endoderm layer (see Fig. 2). t Inner cells also contained AFP in these cases (see Fig. 3).

AFP production by primary and secondary ICMs Primary and secondary ICMs were fixed for the immunoperoxidase reaction for AFP immediately after isolation and at 24 h intervals after culture. Serial sections were cut in each case and the immunoperoxidase reaction was done on each section using anti-AFP antiserum. The morphology of ICMs fixed by the Sainte-Marie technique was not ideal, and individual cells could not be recognized after the immunoperoxidase reaction unless they contained AFP. It was, therefore, not possible to determine the exact proportion of endoderm cells which were positive. A rough estimate was made by a subjective judgment of the proportion of the endoderm layer labelled in each section, and an average of these proportions was made. Since these estimates were not necessarily accurate the ICMs were placed into groups where 0, 5-25, 25-50, 50-75, 75-95 or 100% of endoderm cells were labelled by such a subjective estimate. The results for both primary and secondary ICMs are presented in Table 1. No AFP-positive endoderm cells were observed in the primary endoderm until 48 h after culture, when they appeared in two out of seven ICMs. In a further two ICMs which had no positive cells in the outer layer, AFP-positive cells were observed beneath the primary endoderm (Fig. 2). After 72 h in culture all primary ICMs had at least some AFP-positive cells in the endoderm layer, but the proportion was less than 25 % in all cases. A gradual increase was observed during further culture but the endoderm layer never became 100 % AFP-positive (Table 1, Fig. 2). The maximum labelling of primary endoderm cells was 50%, which was observed in one out of three ICMs after 120 h in culture. Cell differentiation in isolated ICMs 373

50/xm 50 jum

50 jum

Fig. 2. Immunoperoxidase reaction for AFP on sections of primary ICMs after different times in culture: (A) 24 h, (B) 48 h, (C) 72 h, (D) 96 h, (E) 120 h in vitro. AFP-positive cells are present in some outer endoderm cells after 72 h (C, D, E), and in a layer beneath the outer endoderm layer at 48 h (B). The majority of endoderm cells are AFP-negative after 72 h. 374 MARIE DZIADEK

. i

50 Mm

Fig 3 Immunoperoxidase reaction for AFP on sections of secondary ICMs after different times in culture: (A) 24 h, (B) 48 h, (C) 72 h, (D) 72 h, (E) 96 h in vitro. The majority of outer cells are AFP-positive after 72 h (C, D, E) (of primary ICMs, Fig. 2). Some inner cells also contain AFP after 72 h (D, E), probably by adsorption. Cell differentiation in isolated ICMs 375 AFP-positive cells were observed in the majority of secondary ICMs after 24 h in culture, on the seventh equivalent gestation day (Figs. 1 and 3, Table 1). Thus, AFP-positive cells appear in the secondary endoderm layer on the same equivalent gestation day as they appear in primary ICMs (Fig. 1). After 48 h in culture only one secondary ICM had less than 50 % AFP-positive cells in the endoderm layer while six had more than 50 %. After 72 h and 96 h in culture the majority of ICMs had greater than 75 % labelling and five out of seven ICMs were 100% labelled (Table 1, Fig. 3). In these five secondary ICMs cells in the inner core also contained high levels of AFP, most probably by adsorption of AFP secreted by the secondary endoderm cells (Dziadek & Adamson, 1978). In all primary ICMs and secondary ICMs which were not completely AFP- positive, AFP-positive and -negative cells were observed to be randomly distributed in the endoderm layers, with labelled and unlabelled patches inter- mixed (Figs. 2 and 3). The results show that secondary ICMs after 48 h in culture show a significantly higher proportion of AFP-positive cells than primary ICMs after 72 h in culture, indicating that these two layers differ in biochemical activity.

DISCUSSION AFP expression in primary and secondary endoderm layers The results presented in this study show that the endoderm layer which forms on immunosurgically isolated ICMs in vitro produces AFP, a gene product which is expressed in endoderm cells of the intact embryo developing in vivo. When this primary endoderm layer is removed by immunosurgery, a secondary layer forms, in which cells also express this gene product, but the ratio of AFP- positive cells is markedly increased. These results show that AFP synthesis can be initiated in vitro, and is not dependent on the presence of maternal or blastocoelic fluid, the trophectoderm, or the normal structure of the embryo. AFP is a specific gene product of visceral endoderm cells in the early mouse embryo, which appears in a few visceral embryonic endoderm cells on the seventh day of gestation (Dziadek & Adamson, 1978). ICMs are isolated at a stage in blastocyst development equivalent to the fifth day of gestation. AFP- positive cells are first observed in low numbers after 48 h, which is equivalent to the seventh day of gestation. However, although all visceral embryonic endoderm cells in the embryo contain AFP early on the eighth day of gestation, and visceral extra-embryonic endoderm cells do so when isolated from the extra- embryonic ectoderm (Dziadek, 1978), this is not observed in the primary endoderm layer even after 120 h in culture in the absence of trophectoderm cells. If it is assumed that cells which contain AFP are visceral endoderm, this result suggests that the primary endoderm layer is not entirely composed of visceral endoderm cells. By a subjective estimate the AFP-positive cells comprise less than 25 % of the total population in most cases, and reach 50 % only in one 376 MARIE DZIADEK out of three ICMs after 120 h. The majority of cells do not contain AFP, and are likely to be parietal endoderm cells. Previous workers have suggested from morphological observations that the endoderm formed on isolated ICMs is a haphazard mixture of both parietal and visceral endoderm cells (Solter & Knowles, 1975; Strickland et ai, 1976; Pedersen et al, 1977), and the present study provides biochemical evidence in support of this view. When the primary endoderm layer is removed by immunosurgery a distinct second layer is formed within 24 h. After 48 h in culture the majority of cells in the secondary endoderm layer contain AFP, and in some ICMs the entire secondary endoderm is AFP-positive. This is analogous to the embryonic region of the egg cylinder on the eighth day of gestation and is in direct contrast to the primary endoderm layer. Some AFP-negative cells were observed in secondary ICMs even after 72 or 96 h in culture, and these may be parietal endoderm cells. A specific biochemical marker for parietal endoderm cells is clearly necessary for their identification. These results show that visceral endoderm cells are not necessarily derived from the first layer of endoderm cells which forms on the surface of isolated ICMs in vitro, but can be produced from the remaining cells when the first layer is removed. The appearance of AFP-positive cells beneath the primary endoderm layer in two ICMs after 48 h in culture is consistent with this origin.

A model for parietal and visceral endoderm formation The following model can be proposed for the formation of parietal and visceral endoderm cells in isolated ICMs in culture. The first cells to be formed on the outer surface of the ICM are parietal endoderm cells. Visceral endoderm cells are then produced from the ICM ectoderm core, and move to the surface of the ICM amongst the parietal endoderm cells. The presence of the parietal endoderm cells limits the number of visceral endoderm cells which can surface. As the ICM grows in culture, with an increase in surface area, the number of visceral endoderm cells increases, but these are never observed to make up the entire primary endoderm. If surface space is created by the immunosurgical removal of parietal endoderm cells, visceral endoderm cells form a complete layer around the ectoderm. By this model, parietal and visceral endoderm cells do not form from the same stem cell population in the primary endoderm layer, but are formed consecutively from ectoderm cells (Fig. 4). The variability in the number of visceral endoderm cells appearing in the primary endoderm layer may depend on the stage of development at which the ICM was isolated. If some parietal endoderm cells had migrated away from the ICM before isolation (as observed in two out of five blastocysts which were sectioned), fewer may be formed in the primary endoderm, which could result in more visceral endoderm cells appearing in the primary endoderm layer than if developmentally younger ICMs had been isolated. This model for parietal and visceral endoderm formation can be applied to Cell differentiation in isolated ICMs 317

Previous cell lineage (Gardner & Papaioannou, 1975) Parietal endoderm Primitive " ' Trophectoderm Visceral endoderm / Morula >• ICM \ Embryonic^ Ectoderm ectoderm ^ Mesoderm Definitive endoderm Proposed cell lineage Parietal Visceral Trophectoderm endoderm endoderm / / ii / ii / / \\ / i J ^^^ Ectoderm Morula >• ICM >• Primary—$- Secondary ectoderm ectoderm * Mesoderm ^""""^ Definitive endoderm Fig. 4. Diagrammatic representation of alternative cell lineages for parietal and visceral endoderm formation during mouse embryogenesis. ( , Cell lineage; , predicted tissue interactions.) the normal blastocyst in the following way. Parietal endoderm cells which form on the surface of the ICM migrate out over the trophectoderm and are rapidly replaced by visceral endoderm cells from the ICM. Transplantation studies show that cells which contribute to the future visceral yolk-sac endoderm, but not mesoderm or ectoderm derivatives, are already present in the ICM of fifth- day blastocysts (Gardner & Papaioannou, 1975; Gardner & Rossant, 1978). The model gains some indirect support from the morphological studies of Enders, Given & Schlafke (1978) on endoderm formation in mouse and rat blastocysts. Their observations show that prior to primitive endoderm formation the ICM is compacted. After endoderm formation the ICM decompacts and parietal endoderm cells are observed to migrate over the trophectoderm layer. The ICM then recompacts, during which time a layer of cuboidal visceral endoderm cells is seen on the surface of the now elongating ICM. Compaction of cells at the morula stage is thought to be a prerequisite for trophectoderm formation (Ducibella, Albertini, Anderson & Biggers, 1975), and a similar mechanism may operate during endoderm formation. Enders et al. (1978) observed in rat embryos that during the time that parietal endoderm cells doubled in number, the number in the visceral endoderm increased tenfold. If both cell types were derived from the same stem cell population this represents a sudden difference in their cell cycle, with visceral endoderm cells dividing three times as fast as parietal endoderm cells. The alternative proposal that the ectoderm contributes to the visceral endoderm layer is more plausible. The mechanisms by which parietal and visceral endoderm cell types would 378 MARIE DZIADEK form from the ICM in a consecutive manner can only be speculated on at present. Parietal endoderm cells may form in response to an 'outside' position in the ICM. Once formed, parietal endoderm cells may interact with the remaining ICM cells and provide the signal for differentiation of visceral endoderm cells. Likewise, an interaction between visceral endoderm and embryonic ectoderm cells may be necessary to initiate the differentiation of mesoderm and definitive endoderm cells. Thus, early development of the mouse embryo could be viewed as a progressive series of cell-environment and cell-cell interactions.

I would like to thank Dr Eileen Adamson for supplying antisera to AFP, without which this work would not have been possible, and Dr Chris Graham for sound advice and criticism. I was supported by a Flinders University of South Australia Overseas Scholarship.

REFERENCES BODE, V. C. & DZIADEK, M. A. (1979). Plasminogen activator secretion during mouse embryogenesis. Devi Biol. (In the Press.) DUCIBELLA, T., ALBERTINI, D. F., ANDERSON, E. & BIGGERS, J. D. (1975). The preimplantation mammalian embryo: characterization of intercellular junctions and their appearance during development. Devi Biol. 45, 231-250. DULBECCO, R. & VOGT, M. (1954). Plaque formation and isolation of pure lines with polio- myelitis viruses. /. exp. Med. 99, 167-182. DZIADEK, M. (1978). Modulation of alphafetoprotein synthesis in the early postimplantation mouse embryo. J. Embryol. exp. Morph. 46, 135-146. DZIADEK, M. & ADAMSON, E. (1978). Localization and synthesis of alphafetoprotein in postimplantation mouse embryos. J. Embryol. exp. Morph. 43, 289-313. ENDERS, A. C. (1971). The fine structure of the blastocyst. In Biology of the Blastocyst (ed. R. J. Blandau), pp. 71-94. Chicago: University of Chicago Press. ENDERS, A. C, GIVEN, R. L. & SCHLAFKE, S. (1978). Differentiation and migration of endoderm in the rat and mouse at implanatation. Anat. Rec. cxc, 65-78. 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-fetoprotein localization in tissue sections. Int. J. Cancer 7, 198-206. GARDNER, R. L. & PAPAIOANNOU, V. E. (1975). Differentiation in the trophectoderm and inner cell mass. In The Early Development of Mammals (ed. M. Balls and A. E. Wild), pp. 107-132. Cambridge University Press. GARDNER, R. L. & ROSSANT, J. (1978). Investigation of the fate of 4-5 day post-coitum mouse inner cell mass cells by blastocyst injection. J. Embryol. exp. Morph. 52, 141-152. HANDYSIDE, A. H. & BARTON, S. C. (1977). Evaluation of the technique of immunosurgery for the isolation of inner cell masses from mouse blastocysts. /. Embryol. exp. Morph. 37, 217-226. HOGAN, B. & TILLY, R. (1978 a). In vitro development of inner cell masses isolated immunosurgically from mouse blastocysts. I. Inner cell masses from 3-5-day p.c. blastocysts incubated for 24 h before immunosurgery. /. Embryol. exp. Morph. 45, 93-105. HOGAN, B. & TILLY, R. (19786). In vitro development of inner cell masses isolated immunosurgically from mouse blastocysts. II. Inner cell masses from 3-5- to 4-0-day p.c. blastocysts. /. Embryol. exp. Morph. 45, 107-121. NADIJCKA, M. & HILLMAN, N. (1974). Ultrastructural studies of the mouse blastocyst sub- stages. /. Embryol. exp. Morph. 32, 675-695. PEDERSEN, R. A., SPINDLE, A. I. & WJLEY, L. M. (1977). Regeneration of endoderm by ectoderm isolated from mouse blastocysts. Nature, Lond. 270, 435-437. Cell differentiation in isolated ICMs 379 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, 991-1001. SAINTE-MARIE, G. (1962). A paraffin embedding technique for studies employing immuno- fluorescence. /. Histochem. Cytochem. 10, 250-256. SOLTER, D. & KNOWLES, B. B. (1975). Immunosurgery of mouse blastocyst. Proc. natn. Acad. Sci., U.S.A. 72, 5099-5102. 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 embryogenesis. Cell 9, 213-240. WHITTEN, W. K. (1971). Nutrient requirements for the culture of preimplantation embryos in vitro. Advances in Biosciences 6, 129-139.

{Received 27 March 1979, revised 8 May 1979)