Maturation and polarization of the endocytotic system in outside blastomeres during mouse preimplantation development

A major differentiative feature of mouse preimplantation development is the generation, by the blastocyst stage, of a mature, polarized, outer epithelium, the trophectoderm (TE). TE cell polarity is most clearly shown by the presence of junctional complexes, including zonular tight junctions, at the apicolateral margins of apposed cells (Ducibella, Albertini, Anderson & Biggers, 1975; Magnuson, Demsey & Stackpole, 1977) which permit vectorial fluid transport during cavitation and maintain the asymmetry of apical and basolateral membrane compartments. These membrane domains differ in their structure (Calarco & Epstein, 1973; Nadijcka & Hillman, 1974; Johnson & Ziomek, 1982), antigenic properties (Searle & Jenkinson, 1978; Wiley & Eglitis, 1981; Randle, 1982), enzyme constituents (Borland, 1977; Mulnard & Huygens, 1978; Izquierdo, Lopez & Marticorena, 1980; Izquierdo & Ebensperger, 1982; Kaye et al. 1982) and in their relative adhesiveness (Burgoyne & Ducibella, 1977; Kimber, Surani & Barton, 1982) (see also review by Johnson, 1981). Intracellular polarity within the TE is also evident with respect to organelle distribution; mitochondria and cytoplasmic droplets tend to localize close to the basolateral membrane (Wiley & Eglitis, 1981) while secondary lysosomes occur preferentially in the basal cytoplasm (Fleming, Warren, Chisholm & Johnson, 1984). A corresponding cytoplasmic heterogeneity has not been detected within the inner cell mass cells (ICM) of the blastocyst and thus the possession of a polarized phenotype is a characteristic feature of TE.

A stable asymmetry of cellular organization is first evident in 8-cell blastomeres developing during the process of compaction, although regionalization of membrane-associated enzymes (Izquierdo et al. 1980; Izquierdo & Ebensperger, 1982) and of cholesterol organization (Pratt, 1985) has been reported for earlier stages. The polarization of blastomeres at the 8-cell stage is pervasive and involves the redistribution of membrane, cytoskeletal and organelle components. Thus microvilli become localized apically (Ducibella, Ukena, Karnovsky & Anderson, 1977; Handyside, 1980; Reeve & Ziomek, 1981), while focal tight junctions (Ducibella & Anderson, 1975; Magnuson et al. 1977) and functional gap junctions (Lo & Gilula, 1979; Goodall & Johnson, 1982, 1984; McLachlin, Caveney & Kidder, 1983) form at apposed cell surfaces. These structural asymmetries in membrane domains overlie differences in their organization at the molecular level (Pratt, 1985). A corresponding asymmetry within the cytoplasm is evident for endocytotic vesicles (Reeve, 1981a) and cytoskeletal actin (Johnson & Maro, 1984) which polarize apically, while the nucleus occupies a basal position (Reeve & Kelly, 1983). Changes in the distribution of other cellular components including clathrin and various membrane organelles during the 8- and 16-cell stage have been reported by Maro, Johnson, Pickering & Louvard (1985).

The heterogeneity in the cell surface at the 8-cell stage is maintained during division to the 16-cell stage leading to the formation of polar outside and apolar inside cell populations (Johnson & Ziomek, 1981; Reeve, 1981b). The polar and apolar phenotypes thus derived, although retaining a regulative capacity (Ziomek, Johnson & Handyside, 1982; Johnson & Ziomek, 1983), tend to divide conservatively to yield TE and ICM cell types respectively at the blastocyst stage (Ziomek & Johnson, 1982). Thus the process of polarization which occurs over two cell cycles occupies a central role in the generation of cell diversity and the establishment of a functional epithelium (reviewed Johnson, 1985a,b).

In the present paper, the process of TE epithelial development in outside blastomeres is considered in relation to the endocytotic system. Both the structural and molecular heterogeneity within the membrane of a cell in a fully differentiated epithelium and the tendency for internalization of specific molecules to occur from a particular (usually apical) domain (reviewed by Reggio, Coudrier & Louvard, 1982; Rindler, Ivanov, Rodriguez-Boulan & Sabatini, 1982; Sabatini et al. 1983; Herzog, 1984) necessitate a certain rigidity in the spatial organization of the endocytotic system. Such an intracellular ‘response’ to membrane asymmetry is essential both for an efficient processing of extracellular material and, more importantly, to control against potential randomization of surface domains by vectorial membrane traffic (Rodewald & Abrahamson, 1982). It is therefore of interest to establish when such an endocytotic organization becomes established within the mouse early embryo and to determine the time at which endocytotic activity becomes preferentially segregated to the absorptive membrane domain.

Using conventional, quantitative, cytochemical and tracer techniques at the ultrastructural level, we provide evidence that from the 8-cell stage there are incremental advances in the maturation and polarization of the endocytotic apparatus that relate directly to the generation of TE epithelium. These events include the apical polarization of prelysosomal endosomes (endocytotic organelles of Reeve, 1981a,b) at the 8-cell stage, the formation and basal polarization of secondary lysosomes at the 16-cell stage and a progressive increase in apical membrane pinocytotic activity during the period from compaction to cavitation.

For collection of fertilized eggs and of cleavage- and cavitation-stage embryos, female HC-CFLP (Hacking & Churchill Ltd) or MFI (Olac Ltd) mice were superovulated (5i.u. PMS followed 45–48 h later by 5i.u. hCG, Intervet) and mated overnight with HC-CFLP males; a vaginal plug indicated successful mating. Superovulated, unmated mice were used for collection of unfertilized oocytes. Preovulatory oocytes were obtained (a) 48 h post-PMS without hCG injection (germinal vesicle stage) or (b) 6 h post-hCG injection (first metaphase stage; Edwards & Gates, 1959) by piercing individual follicles during ovary dissection.

Recovered oocytes, eggs and embryos were transferred to HEPES-buffered Medium 2+4 mg ml⁻¹ bovine serum albumin (M2+BSA; Fulton & Whittingham, 1978) and cultured where appropriate in Medium 16+4 mg ml⁻¹ BSA (M16+BSA; Whittingham, 1971) under oil in Falcon dishes at 37 °C and 5 % CO₂ in air.

Fertilized eggs and unfertilized oocytes were recovered at 15 h post-hCG and, in the case of presumptive fertilized eggs, cultured in M16+BSA until 20 h post-hCG and examined for the presence of two pronuclei. Two-cell, 4-cell and precompact 8-cell embryos were obtained by flushing oviducts at the late 2-cell stage (47–50 h post-hCG) and cultured in M16+BSA until the required developmental age (4-cell, 50–58h post-hCG; 8-cell 60-70h post-hCG). In some instances, early 2-cell embryos were recovered at 37 h post-hCG.

In experiments where more precise timing of early 8-cell embryos was required, a stock culture of embryos containing four to seven cells was examined at hourly intervals and those embryos that had formed 8-cells since the previous inspection were picked off, cultured separately, and designated Oh 8-cell embryos. Compacted 8-cell embryos were obtained in a similar manner by hourly inspection of stock embryos. Those which had compacted during the previous hour were pooled and designated Oh compacted 8-cell embryos. The number of inspection periods for a given stock of embryos was restricted to avoid inclusion of abnormal or slow developing embryos in subsequent experiments. Thus Oh compacted embryos were aged between 5–9 h from the beginning of the cell cycle.

Embryos at different ages throughout the 16- and 32-cell morula stages were likewise timed from compaction. However, in these experiments, embryos were recovered at the early 8-cell stage (60-63 h post-hCG) with any early compacted embryos removed prior to culturing. 0 h compacted embryos thus derived were cultured for varying times until required. Blastocysts were obtained by culturing morulae recovered from oviducts at 65–72 h post-hCG; for some experiments blastocysts were staged from the onset of cavitation (Chisholm et al. 1985).

Removal of the zona pellucida was achieved by brief exposure (15–20 s) to prewarmed acid Tyrode’s solution containing 4 mg ml⁻¹ polyvinylpyrrolidone (Nicolson, Yanagimachi & Yanagimachi, 1975) followed by extensive washing in M2+BSA.

Embryo decompaction was carried out by incubation in Ca²⁺-free M2 or M16+6mgml⁻¹ BSA for 10–15 min at 37°C.

Routine labelling of endocytotic organelles was achieved by incubation for the specified time in M16+BSA containing 5mgml⁻¹ horseradish peroxidase (HRP; Sigma Type II) at 37°C. Prolonged HRP incubation does not appear to have any detrimental effect on subsequent development (blastocyst formation, blastocyst and ICM cell number, potential to form outgrowths) according to Reeve (1981a).

Late 4-cell (55-60 h post-hCG) and compacted 8-cell embryos were freed from their zonae, and disaggregated to single blastomeres using a flame-polished micropipette following incubation in Ca²⁺-free medium. Small groups (four or five) of 1/4 or 1/8 cells thus derived were transferred into drops of HRP-containing M16+BSA with the cells spaced apart, and returned to culture. At hourly intervals, 1/4 or 1/8 cell stocks were examined for division to 2/8 or 2/16 couplets respectively. Divided cells from each inspection were collected, designated Oh 2/8 or 2/16 couplets, and cultured as individual pairs for varying periods in microdrops of M16+BSA with or without HRP (see below) in Sterilin dishes; for 2/16 couplets, Ca²⁺-free medium was used. 0 h 2/8 couplets were cultured for varying periods between 0 and 8 h whilst 2/16 pairs were cultured for between 0 and 12h. When the culture period was for 6h or less, couplets were maintained throughout in M16+BSA; pairs cultured for longer periods were initially placed in HRP-containing medium and transferred to M16+BSA for the terminal 6h.

At the end of the culture period, pairs were labelled by indirect immunofluorescence for evidence of surface polarity using heat-inactivated rabbit anti-mouse species antiserum (RAMS) diluted 1:15 in M2+BSA for 10min, washed three times in M2+BSA and incubated in fluorescein-conjugated goat anti-rabbit IgG (FITC-GAR; Miles Labs) diluted 1:15 in M2+BSA for 10min. Couplets were subsequently washed again in M2+BSA and fixed for 15 min in 4 % paraformaldehyde in phosphate-buffered saline, and returned to M2+BSA for temporary storage.

HRP localization was visualized using the aminoethylcarbazole (AEC) reaction (Reeve, 1981a,b). Two mg AEC (Sigma) were dissolved in 0·5 ml dimethylformamide to which was added 9·5 ml acetate buffer (pH5·0, 50HIM). The reaction mixture was Millipore filtered and, immediately before use, one drop of freshly prepared 3 % (v/v) H₂O₂ was added. Cells were stained for 20–30 min, washed in M2+BSA and transferred to wells of a tissue-typing slide (Baird & Tatlock) in drops of M2+BSA under oil.

Cells were examined for HRP localization and FITC labelling using a Zeiss Universal photomicroscope fitted with a Zeiss filter set 487709. Photomicrographs were taken with Kodak Tri-X 35 mm film.

In addition to qualitative fine structural studies on the morphology and localization of endocytotic organelles within oocytes and a range of preimplantation stages, quantitative TEM analyses were also carried out on particular stages as follows.

Endosome distribution was assessed within blastomeres from zona-intact early 8-cell embryos (2h postformation), 1h compacted 8-cell embryos and 9h compacted 16-cell embryos (outside blastomeres only). Large-size electron micrographs (standardized magnification ×7000; print size 23×20 cm) of cells sectioned approximately through the centre of the nucleus were used to record endosome number per unit area of apical, basal or total cytoplasm. To avoid misinterpretation, profiles of endosome-like structures smaller than 0·1 μm diameter were not included in the analysis. Apical and basal cytoplasm were separated on each print by a line drawn across the cell, through the centre of the nucleus on an axis parallel to the apical surface or, when this was convex, parallel to a line linking the apical points of intercellular contact. Cells with unusual profiles that did not conform to this method of analysis were omitted. A total of 110 cells (approximately 2 cells per embryo) was analysed; the means obtained for each embryo stage were compared using Student’s t-test.

Estimates of the relative pinocytotic activity occurring at apical and basolateral membrane faces were obtained for the following zona-intact stages: early 8-cell (2h postformation), lh compacted 8-cell, 9h compacted 16-cell; 6h postcavitation blastocyst. Ih and 9h decompacted 8- and 16-cell embryos were also analysed as controls for tracer accessibility (see below).

Embryos at each stage were incubated in M16+BSA containing HRP for 10 or 20 min at 37°C, rapidly washed in M2+BSA (15–20 s) and fixed for electron microscopy; 8-cell stages were also incubated in HRP for 5 min. Decompacted embryos were incubated, and also subsequently washed, in calcium-free medium. Blastocysts were collapsed with a microinjection needle (tip diameter 0·5 μm) 45 min prior to HRP incubation in order to equalize tracer accessibility to both sides of the TE cells.

Unstained ultrasections taken from the midregion of embryos from each group were assessed for the number of HRP-positive pinocytotic vesicles (PVs) occurring at apical and basolateral membrane faces of outside cells. For all groups except decompacted 8- and 16-cell stages, the apical surface was defined as the outward facing region of membrane, free of cell contact. For decompacted stages, apical membrane referred to the region corresponding to the microvillous pole. Analyses were confined to blastomeres sectioned through their nucleus and, to reduce error in determining PV derivation, only those PVs occurring within 0·5μm of the respective membrane face were scored. Data obtained for each cell profile scored were standardized to the length of the relative membrane face (excluding microvilli) derived from measurements made on large-size electron micrographs at a fixed magnification; mean values were finally expressed as an index of pinocytosis (IP) = PV number/10 μm apical or basolateral membrane. PV data for the whole cell profile were also recorded. A total of 271 cells was examined and Student’s t-test was used for statistical comparisons.

Early and compacted 8-cell stages and compacted 16-cell embryos labelled with HRP for specific times prior to fixation and used in the analysis of pinocytotic activity described above were also employed to compare the extent of endosome labelling with time and embryo stage. The proportion of HRP-positive endosomes in apical, basal and total cytoplasm of suitable cell profiles was determined from electron micrographs at a standardized magnification. Apical and basal cytoplasmic zones were as defined in the endosome distribution analysis. A total of 141 cells was examined; mean values were compared using Student’s t-test following arcsin transformation of percentage data (Bishop, 1966).

Three techniques were employed for the detection of lysosomal hydrolases at the ultra-structural level: acid phosphatase, aryl sulphatase and trimetaphosphatase. For each enzyme, late 2-cell, lh compacted 8-cell and mid-expanded blastocyst stages, both with and without zonae, were examined following glutaraldehyde fixation. For the trimetaphosphatase reaction, 12 h postcompaction 16-cell stage morulae were also processed.

The acid phosphatase reaction was carried out according to the modified Gomori procedure described by Lewis & Knight (1977) using sodium β-glycerophosphate (Sigma) as substrate. The reaction was performed both at the conventional pH 5-2 using 0-lM-tris-maleate buffer and at pH4·0 using 0·1 M-acetáte buffer. Embryos were incubated for 1 h at 37°C.

The protocol employed for the detection of aryl sulphatase was that of Hopsu-Havu, Arstila, Helminen & Kalimo (1967), using p-nitrocatechol sulphate (Sigma), as substrate. The reaction was carried out at pH 5-5 and pH 4-0 in acetate buffer; incubations were for 1 h at 37°C.

Trimetaphosphatase localization was carried out according to Doty, Smith, Hand & Oliver (1977) with sodium trimetaphosphate (Sigma) used as substrate. Embryos were incubated at pH 3·9 (acetate buffer), for 1 h at 37°C.

For each procedure, control embryos were incubated in substrate-free medium; mouse liver was used as a positive control.

Embryos were routinely fixed in 3 % glutaraldehyde in 0-lM-cacodylate buffer (pH7-3) for 30 min at room temperature, washed twice in buffer and postfixed in similarly buffered 1 % osmium tetroxide for 30 min. Embryos were washed subsequently in distilled water, en bloc stained with saturated aqueous uranyl acetate for 30min (omitted for HRP-treated embryos), dehydrated in a graded ethanol series and embedded in Spurr’s resin.

In order to label the plasma membrane, embryos (2-cell, compacted 8-cell, blastocysts) without zonae were fixed (both glutaraldehyde and osmium steps) in the presence of 0·1 % ruthenium red (Sigma) according to Luft (1971).

For enzyme localization, HRP-treated embryos were incubated after glutaraldehyde fixation for 30 min at room temperature in 0·1 M Tris-HCl buffer (pH 7-3) containing 0·5 mg ml⁻¹ diaminobenzidine tetrahydrochloride (DAB; Sigma) and 0·03% H₂O₂. An endogenous peroxidase activity was not detectable at the EM level in DAB-reacted embryos not incubated in HRP.

Ultrathin sections were cut on an LKB Ultrotome III, stained with alcoholic uranyl acetate and lead citrate, and viewed in an AEI 6B electron microscope at 80kV. Unstained sections examined at 40 or 60 kV were routinely used for HRP-incubated material.

The morphology, composition and distribution of the endocytotic organelles and the polarity of the processing pathway were found to alter at different periods during preimplantation development. These phases in the maturation of the endocytotic system are described and, where appropriate, quantified according to the developmental stages at which they occur.

In germinal vesicle and first metaphase staged preovulatory oocytes (Fig. 1A), unfertilized oocytes, fertilized eggs, and cleavage stages up to the 4-cell stage (Fig. 1B), the endocytotic organelles occur as discrete vesicular clusters localized almost exclusively in the cortical cytoplasm. Clusters in early cleavage stages are generally smaller than those in oocytes and are found beneath both outward facing and apposed regions of the cell membrane (Fig. 1B). Following incubation in horseradish peroxidase (HRP), these organelles are the only intracellular structures that accumulate tracer (Fig. 1C).

Each endocytotic cluster consists of several spherical or flattened, electron-lucent vacuoles (0·1–0·5μm diameter) with intervening smaller (70 nm), medium dense tubular or vesicular structures (TVS) which are often symmetrically arranged around the vacuoles (Fig. 2A,C). In some clusters, vesicular structures are also found inside the vacuolar compartment (Fig. 2B), thus explaining the classical terminology of these organelles within rat (Sotelo & Porter, 1959; Izquierdo & Vial, 1962; Schlafke & Enders, 1967; Stastna, 1974; Dvorak, 1978) and mouse (Calarco & Brown, 1969) blastomeres as multivesicular bodies. However, for consistency with the modern literature on prelysosomal endocytotic compartments (Helenius, Mellman, Wall & Hubbard, 1983), the term endosome will be used hereafter. In differentiated cells, these organelles represent the initial site of accumulation of internalized substances where, by virtue of their acidic yet hydrolase-free milieu, components are ‘sorted’ prior to further processing (Helenius et al. 1983). A close interrelationship between the three components of the mouse embryo endosome is evident in favourable sections where vacuole and TVS membranes are seen in continuity (Fig. 2C) and a deep invagination is found in the vacuolar membrane indicative of the formation of internalized vesicles (Fig. 2D). In a small proportion of endosome profiles, an electron-dense ‘mat’ is present on one region of the cytoplasmic face of the vacuole membrane which is free of TVS elements (Fig. 2E).

When 2-cell embryos were fixed in the presence of ruthenium red as a marker for cell membrane, endosome clusters were consistently unlabelled (Fig. 2F) and therefore represent true intracellular organelles lacking direct membrane continuity with the cell surface. An identical result was obtained for 8-cell and blastocyst stages.

Three types of organelle are found localized in the cytoplasm surrounding endosome clusters; their occurrence varies according to developmental stage. In oocytes and to a lesser extent at later stages, these comprise spherical multilayered membraneous whorls (0· 2– 0· 5μm diameter; Fig. 3A) which may also contain internalized vesicles (Fig. 3A; inset) suggesting that they are derived from endosomes undergoing senescence. In 2-cell and later staged embryos, endosome clusters commonly colocalize (without direct association) with closely packed granules (0· 2– 0· 4um diameter) containing a homogeneous material and charac-terized by their densely stained apposed membranes (‘jigsaw’ granules of Calarco & Brown (1969); Fig. 2C). The granules appear to derive from cisternae of smooth endoplasmic reticulum during the 2-cell stage (Fig. 3B) and are ultimately autolysed at the morula stage during the formation of secondary lysosomes (see below). Endosome clusters in 4-cell embryos are occasionally associated with electron-dense ‘degradative vacuoles’; these structures are more common in postcompaction stages (Figs 4B arrows, 5B,G) and appear to derive from endosomes.

All three procedures used for the cytochemical detection of acid hydrolases (acid phosphatase, pH4· 0, 5-2; aryl sulphatase, pH4· 0, 5· 5; trimetaphosphatase, pH 3· 9) gave negative results for these developmental stages despite a consistent positive localization for each technique in control mouse liver.

Throughout the 8-cell stage, the morphology of endosome clusters and their association with other organelles is similar to that of earlier stages; endosome distribution, however, alters at compaction. In early 8-cell embryos (2h postformation), endosomes are localized in two regions of the cell; (a) in the cortical cytoplasm and (b) in the perinuclear zone (Fig. 4A). Endosome counts made on electron micrographs of 48 blastomeres from intact 2h 8-cell embryos have shown that approximately 80 % are cortical. The number of endosomes per unit area of apical and basal cytoplasm was likewise determined by analysing electron micrographs (Table 1). No significant difference in endosome localization was found between these two cytoplasmic zones, thus their distribution, as in earlier staged embryos, can be described as apolar.

In lh postcompaction 8-cell embryos, the distribution of endosomes is usually altered from the pattern described above. Endosomes, along with occasional degradative vacuoles and ‘jigsaw’ granules, tend to localize, but not exclusively, in a polar cluster in the apical cytoplasm above the nucleus and below the microvillous pole (Fig. 4B), thus correlating with the distribution pattern of HRP-labelled endocytotic organelles described for these cells by Reeve (1981a). The extent of endosome polarity in compacted 8-cell embryos is variable when whole cells/ embryos are scored for its presence or absence at the light microscope level and has been evaluated at 55 % of embryos examined (Reeve, 1981a). In order to quantify the difference in the relative concentration of endosomes between apical and basal regions of the cell, an analysis of endosome number per unit area of cytoplasm was carried out on blastomeres from intact 1h compacted embryos, comparable with that performed on early 8-cell embryos. The results (Table 1) show that endosome frequency in the apical cytoplasm is approximately twice that in the basal cytoplasm (P< 0-001). Since there was no significant difference in endosome frequency for the whole cell between early and compacted 8-cell stages (Table 1), the process of polarization is likely to result from cytoplasmic relocation rather than by preferential assembly of endosomes in the apical region of the cell.

Recent data on the role of the cytoskeleton in organizing endosome polarity (Fleming, Cannon & Pickering, 1985) support this view.

The endocytotic processing pathway at the 8-cell stage was investigated by HRP labelling. Initially, intact 3h 8-cell embryos were incubated in HRP-containing medium for a prolonged period (5 h) during which time they compacted, in order to label all components of the pathway. Before fixation, embryos was washed several times in M2+BSA to remove exogenous tracer (HRP is a ‘content’ marker and when unconjugated will not bind to membrane) so that all positive sites could reasonably be interpreted as intracellular. By this treatment, HRP was not found within the subzonal space nor between compacted cell membranes thus confirming the adequacy of the washing procedures. HRP was localized primarily in the endosome clusters (Fig. 5A) but was also found (a) within pinocytotic vesicles (PVs; 70 nm diameter) mainly distributed in the cortical cytoplasm close to the cell membrane but also occurring elsewhere, especially in the vicinity of endosome clusters (Fig. 5A), and (b) within degradative vacuoles when present, usually associated with endosome aggregates (Fig. 5B). At higher magnification, HRP within endosomes was restricted to the vacuolar compartment where it coated the inner surface of the membrane and surrounded internalized vesicles, but was not normally found within elements of the tubular and vesicular structures (Fig. 5C, TVS). Comparable labelling patterns have been obtained with anionic ferritin (data not shown).

The time course of the processing pathway was investigated in early (2h postformation) and late (1 h postcompaction) 8-cell embryos incubated in HRP for periods varying between 1–40 min. After 1min incubation, endosomes were unlabelled but infrequent positive pinocytotic vesicles were detected beneath the cell membrane. By 5 min incubation, positive PVs were found both in the vicinity of the membrane and deeper within the cytoplasm close to or fusing with endosomes (Fig. 5D,E), of which a small number within each cluster was labelled. The proportion of endosomes labelled following HRP incubations for 5, 10 and 20 min was compared between early and compacted 8-cell embryos by analysis of electron micrographs (Table 2). For both stages, the proportion of HRP-positive endosomes increased significantly with time, with approximately 70–75 % contain-ing HRP by 20 min incubation. However there was no significant difference between early and compacted embryos, nor between endosomes localized in the apical versus the basal cytoplasm, for either stage. Degradative vacuoles remained unlabelled following incubations for up to 20 min, but a number contained HRP by 30–40 min incubation.

The nature of the processing pathway linking endosomes with degradative vacuoles is obscure since, in contrast to the labelling of endosomes by pinocytotic vesicles described above, HRP-containing vesicles (endosome or cell membrane derived) were not observed to fuse with degradative vacuoles. However, amongst the population of endosomes found either within the apical pole or localized elsewhere, a number was observed with an intermediate morphology showing (a) an increase in size, (b) an accumulation of electron-dense material and (c) a reduction or loss of tubular and vesicular structures (Fig. 5F,G,). Thus, it is likely that degradative vacuoles acquire tracer by a process of endosome transformation.

All techniques used for the localization of lysosomal hydrolases gave negative results in both early and compacted 8-cell blastomeres (Fig. 9D).

The endocytotic processing pathway at earlier stages (oocyte to 4-cell stage) was found not to differ from that described above.

The time at which endosome polarization occurs during the fourth cell cycle was investigated in natural 2/8 couplets cultured from division of 1/4 blastomeres over the period 0–8 h and observed at hourly intervals. In order to correlate this event with surface polarization of microvilli, cells were both preincubated in HRP to label the endosome compartment and stained subsequently by indirect immunofluorescence to label the cell surface; results are shown in Table 3.

The majority of cells aged up to 5 h postdivision were found to be apolar for both HRP localization and surface staining (Table 3; column 3; Fig. 6A). However, an increasing number of cells show apical polarization of endosomes from 4 h postdivision (column 7) whereas a comparable rise in the incidence of surface polarization is not evident until 6h postdivision (column 8). During this period (4–6 h) there is a sizeable population of cells (approximately 25 %) which are only polarized for endosome localization (column 4; Fig. 6B). Following culture for 7–8 h, the incidence of surface polarity exceeds that of endosome polarity (compare columns 7 and 8) and a subpopulation of cells showing only surface polarity is now evident (column 5; Fig. 6D). In the major cell subpopulation, however, both types of polarization are evident (column 6; Fig. 6C). Where both features of polarity are present they colocalize and are almost always opposite to the point of intercellular contact, confirming the observation of Reeve (1981a).

From these results we conclude that endosome polarization usually precedes surface polarization by 1–2 h but that the period over which the former event -occurs for a population of cells may be longer than that required for the establishment of surface polarization.

The preferential localization of endosomes within the apical cytoplasm and below the microvillous pole in compacted 8-cell blastomeres is suggestive of a polarization of the endocytotic pathway to this region of the cell. To determine whether pinocytotic activity occurs preferentially at the apical membrane compared to the basolateral cell surface, the number of pinocytotic vesicles (PVs) (labelled by HRP incubation) localized within 0·5 μm of either of these two domains was ascertained and standardized to a unit length of cell membrane (= index of pinocytosis, IP; see Materials & Methods). The data for early, compacted and decompacted 8-cell embryos are given in Table 4.

Although in cohipacted embryos the index of pinocytosis for apical membrane is greater than that for basolateral membrane following HRP labelling for 5 or 10 min, the difference is not significant. However, when embryos are incubated in HRP for 20 min, by which time the number of membrane-associated PVs has increased, this difference is significant (P < 0·01) and represents a 50 % increase in pinocytotic activity at the apical surface over that at the basolateral surface. Comparative data for early 8-cell embryos following 5, 10 or 20 min incubations show no significant difference in pinocytotic activity between apical and basolateral membranes. Since the data for the whole cell do not differ between early and compacted embryos (for all incubation periods), the elevated level of apical pinocytosis in compacted cells detectable after 20 min incubation is more likely to represent a clustering of membrane regions involved in endocytosis to the apical membrane rather than an increase per se in the number of such sites at the cell apex. The difference in pinocytotic activity between apical and basolateral membranes at compaction is not a reflection of limited accessibility of tracer since a comparable ratio (1-59:1, apical: basolateral IP, 20 min incubation) is obtained when decompacted embryos are analysed. Both apical and basolateral pinocytotic indices are, however, equally reduced in decompacted embryos compared with compacted stages and this reduction is likely to result from the inhibitory effects of low Ca²⁺ medium on endocytosis (Duncan & Lloyd, 1978).

Outside blastomeres from relatively early 16-cell embryos (9 h postcompaction) usually exhibit a polar distribution of endosomes in the apical cytoplasm in a manner comparable with compacted 8-cell stages (Fig. 7). Cells that appear inside possess apolar endosome clusters reminiscent of the condition in early 8-cell embryos (Fig. 7). Analysis of endosome number per unit area of cytoplasm in outside cells (Table 1) demonstrates a significant difference (P< 0·001) in the relative concentration of these organelles between apical and basal zones of the cell with a ratio (2·2:1) which exceeds that obtained for the compacted 8-cell stage. Furthermore, the concentration of endosomes in the total cytoplasm is higher (P<0·01) than in 8-cell stages, indicating either an increase in endosome formation or their differential segregation to outer polar cells at division.

The characteristics of endosome labelling following HRP incubation in outside 16-cell blastomeres did not show differences from the 8-cell stage (Table 2). The proportion of endosomes labelled after 10 and 20 min incubations (61 % and 77 % respectively) was higher, but not significantly so, than the comparable data obtained for the compacted 8-cell stage. Similarly, there was no significant difference between apical and basal populations of endosomes with respect to the percentage labelled for either incubation time.

The relative pinocytotic activity (IP) of apical and basolateral faces of outside 16-cell blastomeres was assessed following HRP labelling for 10 or 20 min using the same procedure employed at the 8-cell stage (Table 5). The IP of apical membrane was significantly higher (P< 0·001) than that of basolateral membrane at both incubation times, with a ratio approximating 2:1. Thus a sharper distinction is evident in endocytotic activity between apical and basolateral surfaces in polar 16-cells than in polar 8-cells, yet the total cell activity for both stages is not significantly different, suggesting that the polarization of the endocytotic pathway is accomplished by a continuous clustering of pre-existing membrane sites involved in endocytosis to the apical surface. In control (decompacted) 16-cell embryos, apical and basolateral IPs are significantly different (P< 0·001) with ratios comparable with those obtained for experimental embryos; thus preferential apical endocytosis is not attributable to limited accessibility of HRP to basolateral surfaces. Interestingly, the inhibitory effect of external calcium deficiency on endocytosis noted in decompacted 8-cell embryos was not apparent at the 16-cell stage. The IP data for trophectoderm cells within mid-expanded blastocysts (Table 5) demonstrate a further increase in the polarization of the endocytotic pathway with the number of profiles of pinocytotic vesicles at the apical membrane exceeding those at the basolateral surface by a ratio greater than 3:1 for both incubation times.

During the later period of the 16-cell stage, and continuing into the 32-cell stage, a further maturation of the endocytotic apparatus is evident in both outside and enclosed cells. These events culminate in the formation of a pool of secondary lysosomes which, in outside blastomeres, are polarized in the basal region of the cell. A description of these changes will precede the data quantifying their time of occurrence.

Outside blastomeres from embryos aged 9 – 25 h postcompaction (early/mid 16-cell to mid 32-cell stages) display a reorganization in the distribution of several cytoplasmic organelles which become localized in the basal cytoplasm (Fig. 8A). These structures may include mitochondria, endosome clusters, cytoplasmic (‘jigsaw’) granules, crystalline arrays, lipid droplets, Golgi bodies and degradative vacuoles (Fig. 8B). The basal aggregates of cytoplasmic granules (which in early cleavage stages colocalize with endosome clusters and are characterized by electron-dense membranes, see Fig. 2C) undergo a process of autolysis which involves (a) their isolation from surrounding cytoplasm resulting in the formation of artifactual lucent spaces and (b) the breakdown of their limiting membranes and the progressive increase in electron density of their contents (Fig. 8B). The end product of granule autolysis is a heterogeneous, electron-dense structure (secondary lysosome, see below) of comparable morphology to the degradative vacuoles already described. During the 16- to 32-cell stage these accumulate progressively in the basal cytoplasm of outside cells (Fig. 9A), but occur in a more random distribution within enclosed cells. An apical pole of endosomes is how-ever retained in outside blastomeres (Fig. 9A), although it is normally of smaller size compared with the early 16-cell condition. In cavitated embryos, the apicobasal disposition of endocytotic organelles in outside (trophectoderm) cells is unchanged; electron-dense secondary lysosomes are evident particularly within basal TE processes which cover the juxtacoelic face of the ICM (Fleming et al. 1984).

Cytochemical examination of 16-cell morulae (12 h postcompaction) and midexpanded blastocysts for lysosomes failed to detect unambiguous acid phosphatase or aryl sulphatase activity but a positive trimetaphosphatase reaction (negative in controls) was obtained consistently in the growing population of basal electron-dense vacuoles, (Fig. 9B), including those located within juxtacoelic processes (Fig. 9C), thus demonstrating their secondary lysosome status. Endosomes and intact aggregates of cytoplasmic granules were non-reactive for trimetaphosphatase.

When late 16-cell morulae are incubated in HRP, the tracer becomes localized within the forming population of basal secondary lysosomes after 30 – 40 min (Fig. 9E). (See Fleming et al. (1984) for evidence of HRP localization within the more substantial pole of basal lysosomes in early blastocysts.) For those embryos fixed directly from the incubation medium, HRP is retained in regions of cell contact (Fig. 9E) thus demonstrating the paracellular movement of the tracer and confirming that lysosome polarization precedes zonular tight junction formation.

The timing of the events leading to the formation and basal polarization of secondary lysosomes within outside blastomeres was examined in ultrasections of intact 16- to 32-cell morulae timed from compaction (range 9 – 25 h) prior to fixation. Cells (sectioned through the nucleus) were scored for the presence or absence of a basal pole of (a) granules undergoing autolysis and (b) secondary lysosomes (Fig. 10). Basal localization of lysosomes was first detected in a minority of outside cells at 10 h postcompaction and reached a maximum of 75 – 80 % by 17 h postcompaction; the proportion of cells displaying granule degradation began to decline from 21 h postcompaction.

A more precise timing of basal localization of secondary lysosomes in relation to the fifth cell cycle was determined by experiments on in vitro cultured cell pairs. Blastomeres derived from compacted 8-cell embryos were cultured individually in the presence of HRP to label the endocytotic organelles and examined hourly for division to 2/16 couplets. These were cultured for varying periods (0 – 12 h) during the fifth cell cycle in Ca²⁺-free medium to prevent intercellular flattening and envelopment (Ziomek & Johnson, 1981) and surface labelled for the detection of the microvillous pole prior to fixation, HRP development and examination. Inmost (83 %) couplets, one cell was clearly larger than its companion. Since larger cells from natural 2/16 couplets have been shown to represent the in vitro equivalent of polar outside cells in 95 % of cases (Johnson & Ziomek, 1981), such cells with polar surface labelling were scored for the distribution of HRP within the cytoplasm (Table 6). However, the orientation of the division plane of 1/8 cells in culture is variable and may bisect the surface pole thereby generating two polar daughter cells (Fig. 11A). Thus the position of the inherited surface pole may bear little relation to the point of intercellular contact. HRP distribution in larger polar cells was therefore scored with the microvillous pole acting as a reference point for the apex of the cell.

In the majority of larger polar cells, HRP sites were preferentially localized within one region of the cytoplasm (focal, Table 6, columns 4 – 7). With few exceptions, such cells displayed HRP polarized either in the apical cytoplasm corresponding to the pole of endosomes (column 4, Fig. 11C) or in the basal cytoplasm corresponding to the pole of secondary lysosomes (column 5, Fig. 11D); only in a minority of cases were focal HRP sites distributed elsewhere (columns 6,7). In the Oh culture group however, a substantial proportion (44%) of larger cells with a polar surface phenotype displayed an apolar or ambiguous disposition of HRP (Fig. 11B) indicating that the spatial organization of the endocytotic system is not maintained continuously during cytokinesis but becomes re-established subsequently.

Comparison between columns 4 and 5 of Table 6 shows that during the 12 h culture period, there is a shift with time in the predominant form of HRP localization from the apical cytoplasm (endosomes) to the basal cytoplasm (secondary lysosomes), with the transition to the latter pattern occurring by 9h postdivision. These results are therefore consistent with those obtained from sectioned material (Fig. 10) considering that division to the 16-cell stage is likely to occur from 2 to 4 h postcompaction.

Late 2-cell embryos (48 h post-hCG) were cultured up to the late 8-cell stage (lh postcompaction; 70 – 72h post-hCG), the early 16-cell stage (6h postcompaction; 75 – 78 h post-hCG) or the late 16-/early 32-cell stage (15 h post-compaction; 84 – 88 h post-hCG) in M16 containing an elevated concentration of BSA (20 mg ml⁻¹) to examine whether the timing of basal polarization of secondary lysosomes in outside cells, normally occurring during the late morula stage, could be influenced by the relative amount of exogeneous protein endocytosed during cleavage. At the two earlier stages, there was a marked increase in the number of electron-dense vacuoles found within the cytoplasm compared with control embryos cultured in M16+4mgml⁻¹ BSA. However, these were consistently situated in the apical cytoplasm associated with the endosome poles (Fig. 12); only in the later staged embryos, including controls, were secondary lysosomes localized basally. Thus the timing of the maturation of the endocytotic pathway appears to be relatively independent of environmental factors and therefore is likely to represent an integral stage in the temporally regulated developmental programme leading to the formation of a polarized epithelium.

Our studies on the early developmental stages of the mouse embryo have shown that during cleavage the endocytotic system undergoes two accompanying cellular processes, namely (a) the maturation of the lysosomal processing pathway for exogenous material, and (b) in outside blastomeres, the polarization of endocytotic components along the apicobasal axis of the cell. Both events proceed in a temporal sequence in outside cells with the result that three distinct phases or increments can be recognized in the generation of the trophectoderm (TE) epithelium. These stages may be defined as (a) the apolar condition (early cleavage up to and including the early 8-cell stage) characterized by the symmetrical organization of an immature endocytotic apparatus; (b) the primary polarized condition (mid 8-cell to mid 16-cell stage) in which non-lysosomal endocytotic organelles localize predominantly in the apical cytoplasm to coincide with preferential endocytotic activity from the apical membrane; and (c) the secondary polarized condition (late 16-cell stage onwards) characterized by the manufacturem and basal polarization of secondary lysosomes to complete the endocytotic pathway.

The transition between phases (a) and (b) and the onset of endocytotic polarity during the 8-cell stage provides further evidence that lineage differentiation in the preimplantation embryo is a consequence of cellular events initiated at the time of compaction (Johnson, 1985a). The endocytotic organelles which polarize in the apical cytoplasm correspond to those described by Reeve (1981a) who first demonstrated the polarity of internalized HRP sites at compaction. These have been designated endosomes since they represent the primary intracellular collecting site for exogenous tracers and show no evidence of acid hydrolase activity (Helenius et al. 1983). Their morphology, characterized by associated tubular/vesicular structures (TVS) and, in some cases, internalized vesicles (multi-vesicular bodies), is also consistent with endosomal organelles described else-where (Helenius et al. 1983). It is of interest that the TVS surrounding endosomes provided little evidence of HRP labelling; this region of the complex has been shown in studies on the TE epithelium to originate from the endosome vacuole and to represent the initial stage of either recycling or transcellular pathways for internalized membrane and therefore generally do not incorporate tracers such as HRP which remain unassociated with membrane (unpublished observations).

The data on endosome distribution suggest that their apical polarization is a progressive rather than a rapid cellular event since (a) their mean concentration within the apical cytoplasm is increased at the early 16-cell stage from that at the compacted 8-cell stage (Table 1) and (b) the incidence of cells with apically polarized HRP sites (endosome pole) within cultured 2/8 or 2/16 (larger, surface polar cells) couplets increased steadily from the mid 8-cell stage (Table 3), reaching a peak at 3 h postdivision to the 16-cell stage (Table 6). An increase in the incidence of restricted HRP staining between polarized 8- and 16-cells has also been recorded by Reeve (1981a,b). This pattern is in contrast with the more synchronous appearance of surface polarity during the 8-cell stage which, although it is initiated somewhat later than endosome polarity, becomes established within the cell population relatively quickly (Table 3). Thus it is plausible that once the inductive signal to polarize is ‘received’ by the cell, the mechanisms involved in redistributing surface membrane and endosomal components remain distinct.

At present, the mechanism controlling surface reorganization has not been determined but recent evidence has indicated that endosome relocation is regulated by the cytoskeleton since polarization is sensitive to both cytochalasin D and colcemid treatments (Fleming et al. 1985). Furthermore, cytoplasmic actin (Johnson & Maro, 1984) and clathrin (Maro et al. 1985), a molecule intimately involved in endocytosis, have been shown to polarize apically at a similar time to endosomes, and before the appearance of surface polarity, during the 8-cell stage. Whether these represent separate events or are the manifestation of a single cytological process remains to be seen. The progressive nature of endosome polarization during the early morula period may imply a continuous process of cytoskeletal redistribution to reinforce or further stabilize the newly-established axis of polarity. Polarized HRP sites, as well as polar actin and clathrin (Johnson & Maro, 1984; Maro et al. 1985), are not preserved during division to the 16-cell stage (Reeve, 1981b) and are evident at a reduced frequency in larger, surface polar cells immediately postdivision before their restoration to beneath the surface pole later in the cell cycle (Table 6). Since surface polarity is evident throughout the same period, it is possible that cytocortical foci direct the reassembly of a polarized internal organization, probably via cytoskeletal elements (Johnson & Maro, 1985). One route by which the cytocortex might influence cytoplasmic organization derives from the current observation that pinocytotic activity is polarized to the apical surface during the 8- and 16-cell stages in a progressive manner (compare Tables 4 and 5), thus typifying the condition found in fully differentiated epithelia (Herzog, 1984). The level of heterogeneous pinocytotic activity may therefore correlate directly with the degree of organelle polarization since it is likely to be limited by the availability of membrane recycling pathways provided by endosomes.

The second phase of endocytotic development initiated in outside cells at the late 16-cell stage is characterized by the manufacture and basal polarization of trimetaphosphatase-positive secondary lysosomes which accumulate HRP after prolonged incubation. Several organelle types including Golgi bodies and cytoplasmic granules were found to migrate to the basal zone during this period indicating that an extensive reorganization of cytoplasmic contents by the cyto-skeleton has again taken place. These observations are supported by immuno-cytochemical evidence demonstrating a basal concentration of lysosomal and Golgi antigens at this time (Maro et al. 1985). Since the primary axis of cell polarity is preserved during this period of organelle redistribution, the formation and restricted localization of secondary lysosomes can be viewed as a maturation stage rather than a differentiative event in the generation of the polar cell lineage. In this context, the disposition of organelles within precursor TE cells resembles more closely the pattern observed in fully differentiated cells where Golgi elements and secondary lysosomes are positioned under the control of the cytoskeleton, along the axis of polarity (Kupfer, Dennert & Singer, 1983; Collot, Louvard & Singer, 1984).

The morphological and tracer observations indicate that basal secondary lysosomes are derived from heterogeneous processes including (a) autolytic digestion of specific cytoplasmic components (e.g. ‘jigsaw’ granules) and (b) endocytotic processing, probably via the transformation of endosomes according to the maturation model described by Helenius et al. (1983). The substructure of secondary lysosomes is comparable with that of degradative vacuoles which occur infrequently but progressively during earlier cleavage stages. However, degradative vacuoles were not found to exhibit hydrolytic activity and hence their endocytotic status (lysosomal?) cannot be classified reliably. A negative cyto-chemical result should be treated with caution since the correct substrate and optimal pH may not have been achieved although three techniques were applied at varying pH values. Indeed previous optical cytochemical analyses on mouse early embryos have recorded a minimal reactivity for acid hydrolases prior to the 16-cell stage (Mulnard, 1965; Solter, Damjanov & Skreb, 1973), and, in an ultrastructural study, hydrolases were first detectable at a low level in the 8-cell embryo and increased thereafter up to the blastocyst stage (Vorbrodt, Konwinski, Solter & Koprowski, 1976). Thus on the basis of cytochemical data alone, we cannot infer that degradative vacuoles are entirely deficient in hydrolase enzymes and hence that the maturation and autolytic processes observed during the late 16-cell stage represent the earliest occurrence of lysosomal activity in the developing embryo. However, despite this qualification, we may at least conclude that at the morula stage there is a substantial upsurge in secondary lysosome production which also includes a qualitative change in that trimetaphosphatase activity is evident for the first time. The mobilization of Golgi elements and their association with organelles undergoing autolysis may signify the localized production of primary lysosomes. Their fusion with surrounding structures including degradative vacuoles would help to explain both the rapid onset of hydrolytic activity detected in this region of the cell and the requirement for basal polarization to congregate target organelles.

Our data are consistent with earlier studies conducted on other mammalian species (rat, rabbit) where endocytotic activity (extent of internalization of tracers) has been shown to increase progressively during cleavage (Schlafke & Enders, 1973; Hastings & Enders, 1974). Although we have interpreted the maturational changes in the mouse endocytotic system as an integral part of the process of cell polarization, and hence of tissue diversification (Johnson, 1985b), it is more difficult to evaluate a physiological role for these changes, other than in general terms. Biochemical data on total protein content and protein synthetic activity during mouse preimplantation development indicate a dependence on a declining store of maternal proteins for metabolic events until the early morula stage. Subsequently, there is a dramatic rise both in synthesis (six to eight fold) and in total protein by the blastocyst stage (Brinster, Wiebold & Brunner, 1976; Abreu & Brinster, 1978; Sellens, Stein & Sherman, 1981). An increase in endocytosis during cleavage, polarized to the outward facing surface of the compacted embryo and coincident with the emergence of definitive secondary lysosomes, suggests that exogenous proteins may function to a significant extent as a substrate for the recorded upsurge in synthetic activity in the late morula.

We wish to thank Dr M. H. Johnson and our research colleagues for their helpful discussions and criticisms of the manuscript. Our thanks also to Tim Crane and Roger Liles for photographic reproductions and Caroline Hunt, Anne Wright and Fiona Symington for preparing the manuscript. This work was supported by grants from the M.R.C. to Dr M. H. Johnson.