ABSTRACT
Fragments of human foetal organs and blood at 5 –11 weeks of postfertilization development were cultured in radioactive protein precursors. The secreted products were characterized by immunoprecipitation, and by measuring the mobility of the immunoprecipitates on polyacrylamide gels. It was found that secondary human yolk sacks secreted apolipoproteins A1 and B. The work of previous authors on the synthesis of other serum products by this organ and by the foetal liver and by the foetal intestines was confirmed. Within the yolk sack, the endoderm, the blood cells, and the outside epithelium reacted with antibodies against apolipoprotein Al and transferrin. By metabolic labelling of umbilical cord blood, it was found that blood did not secrete apolipoproteins Al and B. Blood cells could therefore not be a source of these secreted products.
INTRODUCTION
The rapid metabolism of human foetal testes and adrenals depends on a supply of lipids which must be delivered to the cell surface bound to apolipoproteins (Carr et al. 1980; Carr et al. 1983). Further, it is known that complexes of lipid and apolipoprotein sustain the rapid growth of some human-teratoma-derived cells which resemble early human embryonic stem cells (Engstrom, Rees & Heath, 1985). We have studied the potential sources of apolipoproteins in early human development, since an abundant intrinsic supply of these lipid carrier molecules may be essential for human embryogenesis.
In the adult, the liver and the intestines synthesize almost all the apolipoproteins which circulate in the blood (rat: Wu & Windmueller, 1979). The same organs synthesize these proteins in the 16-to 22-week postfertilization human foetus (Zannis, Kurnit & Breslow, 1982), and there is also evidence that lipoproteins are synthesized much earlier, at 29 days postfertilization (Gitlin & Biasucci, 1969). We have investigated the synthesis of these molecules in the first 5 to 11 weeks of postfertilization human development, identifying some of the individual apolipoprotein molecules, and concentrating our studies on the secondary yolk sack, because in the mouse embryo the analogous visceral yolk sack is the earliest intrinsic source of apolipoprotein (Shi & Heath, 1984; Meehan et al. 1984).
In humans, the inner cell mass apparently gives rise to the endoderm of the primary yolk sack during the fifth day of postfertilization development. The endoderm of the primary yolk sack, which lies beside the epiblast, seems to persist and form the inner lining of the secondary yolk sack, which is a coherent structure by the 14th day after fertilization (sequential histology: Luckett, 1978). The endoderm of the secondary yolk sack has all the features of a secretory tissue (reviewed by Branca, 1913, and Gonzalez-Crussi, 1979), its cytoplasm is heavily stained by antibodies against prealbumin, albumin, α1-antitrypsin, alphafoeto-protein, and transferrin (Albrechtsen, Wewer & Wimberley, 1980; Jacobsen, Jacobsen & Henriksen, 1981). Consequently, it is a likely source of the variety of serum proteins which the human yolk sack is already known to synthesize in common with the foetal liver and intestines (Gitlin & Perricelli, 1970; Gitlin, Perricelli & Gitlin, 1972).
MATERIALS AND METHODS
1. Foetal samples
Foetal organs were collected exactly as described previously (Thompson et al. 1984). The foetal age was estimated from the date of the reported last menstrual period (LMP). In some cases, the date of the LMP was reported to lie within the span of a particular week; the mid point of this week was taken as the LMP date. In rare instances, no date was reported, and the size of the limbs was used to establish that the age of the embryo fell within the 5-to 11-week postfertilization range included in this study; a question mark is used to indicate the age of these samples (O’Rahilly & Gardner, 1975). In all cases, foetal age is expressed as days postfertilization.
2. Tissue fixation and processing
For histology, the organs were fixed at between 30min and 3 h after foetal aspiration. Routine histology was performed on formal-saline-fixed material, and this procedure was used on all liver samples to avoid confusion with clotted blood. For antigen localization, the organs were fixed for 12 –24h at 5°C in 96% (v/v) aqueous ethanol:glacial acetic acid, 99:1 (Engelhardt’s 1971 modification of Sainte-Marie’s fixative; Sainte-Marie, 1962; Engelhardt, Goussev, Shipova & Abelev, 1971). The tissues were dehydrated through a graded series of alcohols and embedded in Paraplast. For scanning electron microscopy, the organs were fixed for 3 h at 4°C with 2 ·5 % (w/v) glutaraldehyde in 0 ·05 M-cacodylate buffer containing, 5 % (w/v) sucrose at pH 7 ·4. They were washed in several changes of buffer with sucrose alone over 20 h at 4°C, postfixed with 1 % (w/v) osmium tetroxide in buffer for 1 h at 4°C, and dehydrated in a series of ethanols. Following transfer to acetone, they were processed with a Samdri critical-point drier with liquid CO2. They were sputter coated with gold, and examined in a Philips SEM500 microscope.
3. Antibodies and antigens
Antibodies were used to both immunoprecipitate metabolically labelled proteins, and to localize antigens on tissue sections. We purchased rabbit antibodies to the following antigens: apolipoprotein A1 (Seward Laboratories, UAC, Blackfriars Rd, London SEI); human apolipoproteins A1 & B, human transferrin, human albumin, human pre-albumin, and human alphafoetoprotein (Behring, Hoechst House, Salisbury Rd, Hounslow, Middx). Sheep antirabbit IgG conjugated to peroxidase was obtained from Serotec Ltd (Station Rd, Blackthorn, Bicester, Oxon OX6 OTP). Rabbit antibodies against human fibronectin were a gift from Dr D. Turner. Rabbit antibodies to human low-density lipoproteins (LDL) were prepared here (Shi & Heath, 1984).
To test the specificity of the reaction of some of the antibodies with antigens on tissue sections, the antibodies were absorbed with various antigens. These included human albumin and transferrin (Behring), and human apolipoprotein A1 which was purified according to the following protocol. Freshly collected human venous blood was adjusted to a density of 1 ·063 gm/ml with solid KBr (Hatch & Lees, 1968), and it was centrifuged at 105 000g in a 45Ti rotor attached to a Beckman L8 for 20 –22h at 12 –15°C (Havel, Eder & Bragdon, 1955). The bottom fraction was pooled, the density raised to 1 ·125 gm/ml with solid KBr, and this was spun for 18 h at 173000g. The top fraction from this spin was dialysed against PBS (solution A of Dulbecco & Vogt, 1954) for 2 –3 days at 5 °C in a Spectrapor membrane (cut off Mr 2000), and delipidized in 25 vols of ethanol:ether (3:1) for 16 –24 h at room temperature (Chapman, Mills & Ledford, 1975). The precipitate was washed in the same ethanol:ether mixture, and then stored at –20°C until use. This material was run on a 12 ·5% (w/v) polyacrylamide gel, using the method of Laemmli (1970). Following staining with Coomassie brilliant blue, the gel was scanned, and the area under visible peaks was calculated by weighing cut outs of the peak profiles. Between 89 and 96 % (n = 5) of the absorbance was in a single peak with a mobility of 28 ×103, which is the correct relative molecular mass (Mr) for apolipoprotein A1; in addition there was a minor peak of absorbance in the region of albumin on the gel, which amounted to 1 –2% (n = 5) of the absorbance in peaks on these tracks. The commercial preparations of transferrin and albumin were analysed similarly, and the proportion of the absorbance at the correct Mr in these two samples amounted to 88 –93 % (n = 3) and 77 –87% (n = 3) respectively; in neither case was there any absorbance in the apolipoprotein Al region in these preparations.
4. Metabolic labelling
Labelling was begun within 3 –5 h of foetal aspiration. Despite this inevitable delay, we know that cells in most of the eye specimens from the same samples were viable (Hyldahl, 1984), and we only excluded two samples from our reported results on grounds of low total metabolic labelling (two gut fragments weighing 2 and 11mgms). Most foetal organs were torn into fragments of approximately 5 mm3. The stalk was removed from yolk sack samples, which were torn open but otherwise left intact. In most cases, gut samples were cut open along their length. The samples were weighed after excess moisture had been removed by dabbing on filter paper. The weights of the organ fragments are recorded in the legend to Table 1. In most experiments, the individual samples were placed in 1 ml of serum-less medium containing transferrin, high-density lipoproteins (HDL), and low-density lipoproteins (LDL) in a basal medium of Dulbecco’s modified Eagles medium diluted 100-fold with methionine-free minimal essential medium (ECM medium of Heath & Deller, 1983; basal media from Gibco Europe, Paisley, Scotland). The medium was supplemented with 50 –100 μCi of [35S]methionine (Amersham International pic, Amersham, U.K.; specific activity 1310Ci/mmole), and the tissue incubated in this medium for 16 h, at 37°C in an humid mixture of 5 %(v/v) CO2 in air.
Samples of foetal blood were obtained by blowing PBS through the arteries and vein of the umbilical cord. To reduce the possibility of cross contamination with cells which might fall off the allantois and the midgut, the umbilical cord was first severed below the region which contains these organs. The blood cells were pelleted by centrifugation, and taken up in 0 ·8 –1 ·5ml of minimal essential medium (Gibco Europe) containing l/100th of the normal methionine concentration, 0 ·5% (v/v) FCS, and 100 μCi/ml of [35S]methionine. At the end of 20 h incubation, the cells were pelleted by centrifugation, and counted with a haemocytometer. These blood samples contained between 2 –7 ×106 cells. Subsequently, the culture medium was centrifuged at 15000g for 15 min, and the supernatant stored at –70°C before immunoprecipitation. In a few experiments, the organ fragments were labelled in [3H]leucine exactly as described in Thompson et al. 1984.
Radioactive secreted proteins were immunoprecipitated from the culture medium as described by Shi & Heath, 1984. Briefly, between 1/5 and 1/10 of the medium was incubated for 60 min at 4°C with protein A-Sepharose CL-4B (PAS, Pharmacia Ltd, Midsummer Boulevard, Milton Keynes, Bucks MK9 3HP, U.K.), which had been precomplexed with a particular antibody. The PAS:antibody:antigen complexes were removed by centrifugation and the incubation repeated with a second aliquot of PAS.antibody. The beads from the two incubations were combined, and pelleted through 10% (w/v) sucrose in PBS which overlaid a cushion of 20% (w/v) sucrose in PBS, at 15000g for 15 min. The pellets were washed once with 0 ·1% (w/v) bovine serum albumin, 0 ·1% (w/v) sodium dodecyl sulphate in PBS, and then boiled for 5 min in Laemmli sample buffer (1970), which contained 50mM-dithiothreitol.
Samples were subjected to one-dimensional slab gel electrophoresis on a 12 ·5% (w/v) polyacrylamide gel containing 0 ·1 % sodium dodecyl sulphate, using the tris buffer system of Laemmli. After electrophoresis, the gels were processed for fluorography according to Bonner & Laskey (1974). The dried gels were exposed to Fuji RX X-ray film at –70 °C for two weeks before development in Kodak DX 90.
5. Immunolocalization
For immunolocalization, 8 μm sections were attached with diluted chick egg albumin to each of the four wells of a teflon-coated slide (C. A. Hendley Essex Ltd, Oakwood Industrial Estate, Laughton, Essex IG10 3TZ, U.K.), and stored at –20°C until use. They were dewaxed in toluene, rehydrated and taken to 70% (v/v) aqueous ethanol. Control experiments, using the components of the peroxidase reaction alone, showed that endogenous peroxidase activity was inhibited with treatment for 30 min with 3 % (w/v) hydrogen peroxide in methanol. All sections were subsequently taken through this procedure. The slides were washed with 70% (v/v) aqueous ethanol for 10min, air dried, washed in PBS, and given three 10min rinses in either 0 ·5 % (w/v) bovine serum albumin or 0 ·5 % (w/v) gelatin in PBS. Diluted 50 μl aliquots of the primary antibody were placed in the wells in an humid atmosphere for 1 h at room temperature (R.T.), and the PBS and 0 ·5 % BSA in PBS rinses repeated as above. The second antibody was then added and incubation conducted as above. After a PBS rinse, the sections were incubated for two 10min periods in aqueous 10mM-Tris, 0 ·15M-NaCl at pH 7 ·3 (TBS). The peroxidase reaction was developed for about 30s in 0 ·75% (w/v) diaminobenzidine, 0 ·03% (w/v) hydrogen peroxide, 10mM-imidazole in TBS. The reaction was terminated by a rinse in TBS, and the sections were dehydrated through a graded series of ethanols, cleared in toluene, and mounted in DPX. To demonstrate that the second layer antibody did not react with endogenous IgG, some sections were treated with the second antibody alone after the first antibody treatment had been replaced by incubation in 0 ·5 % bovine serum albumin in PBS. No brown peroxidase reaction product developed on these sections. The specificity of the primary antibodies was checked by absorbing each (except the anti-AFP antibody) in separate tests with each of the other antigens at concentrations which were sufficient to completely inhibit the reaction of the homologous antibody. Excess antigen was added to the antibody for 30 min at 37 °C, and the complexes spun down by centrifugation before proceeding to use the supernatant in the standard protocol as above. In each case, a particular antibody reaction was inhibited by absorption with its known target antigen, but the reaction was not inhibited by the other antigens used in this study.
We also attempted to elute and dilute any of the antigens which may have absorbed to cells in the intact embryo. Before reaction with the antibodies against apolipoprotein Al and transferrin, foetal fragments were incubated for 19 h in alpha medium lacking nucleosides and deoxynucleosides (Stanners, Eliceiri & Green, 1971; Gibco Europe), which contained 1% (w/v) bovine serum albumin. Before reaction with the antibodies against AFP and albumin, foetal fragments were cultured in the medium which was used for metabolic labelling; this lacked these molecules (see above).
RESULTS
We have confirmed previous anatomical descriptions of the human secondary yolk sack, and follow older authors in describing the large granulated cells which line the inside of the yolk sack as endoderm cells (Figs 1, 3). As Hesseldahl & Larsen (1969) and Hoyes (1969) observed during the 5th-8th weeks of postfertilization development, the layer of endoderm cells is continuous with columns of similar large cells which extend through the mesenchyme to touch the base of the outside epithelial layer of the sack (samples aged 35, 42, and 50 days postfertilization). These columns appear to be associated with pits which indent the endoderm surface (Fig. 1, arrow and Fig. 3), and they are reminiscent of the pits which are seen on the external surface of the mouse visceral yolk sack endoderm (Hogan & Newman, 1984). In older secondary yolk sacks (samples aged 61 and 81 days postfertilization), the endoderm cells form a simpler columnar epithelium which is completely separated from the outer epithelial layer by the mesenchyme and blood vessels. We also examined secondary yolk sack stalks and can confirm that the endoderm cells extend as a partially closed tube down long regions of the stalk.
Composite scanning electron microscope view of a fractured fragment of the human secondary yolk sack, at 50 days postfertilization. The inner endoderm layer is to the right, and it is possible to see regions where the large (endoderm) cells in this layer make contact with the outside epithelial cells to the left (arrow). These regions of contact appear to be associated with pits on the inner endoderm surface (arrow heads). Blood vessels in the middle mesenchyme layer are marked with asterisks (×200).
Composite scanning electron microscope view of a fractured fragment of the human secondary yolk sack, at 50 days postfertilization. The inner endoderm layer is to the right, and it is possible to see regions where the large (endoderm) cells in this layer make contact with the outside epithelial cells to the left (arrow). These regions of contact appear to be associated with pits on the inner endoderm surface (arrow heads). Blood vessels in the middle mesenchyme layer are marked with asterisks (×200).
Whole view of a 44-day postfertilization secondary yolk sack, showing an unusually swollen stalk (s), and blood vessels in the stalk and in the sack (b). The sack has split open. Scale bar equals 1 mm.
The radiolabelled secreted products of a variety of foetal organs were examined, and the results from weighed samples labelled with [35S]methionine are summarized in Table 1. Human secondary yolk sack, foetal gut, and foetal liver each synthesized apolipoproteins. In contrast, there was no precipitation of radioactive material by antibodies to apolipoprotein Al and to LDL when these were tested on the medium over cultured stomach, trophoblast, adrenal, kidney and brain.
The anti-LDL antibody precipitated major labelled molecules with apparent relative molecular masses of 28 ×103, and >200 ×103. The mobilities of these proteins correspond with the respective known mobilities of apolipoproteins Al, and B (Fig. 5, lane D). The identity of the 28 ×103 and >200 ×103 molecules was confirmed by their precipitation by antibodies which had been raised against these individual apolipoprotein species (Fig. 6, lanes D and I; apolipoprotein B results not shown). From Table 1 it can be seen that we were rarely able to immunoprecipitate apolipoprotein Al from the culture medium with the antiapolipoprotein Al antibody; we attribute this difficulty to its low avidity in comparison with the anti-LDL antiserum. Both antibodies immunoprecipitated molecules with the mobility of apolipoproteins A1 and B from liver culture medium (Fig. 6, lanes D and E). A similar result was obtained with embryonic gut (Fig. 6, lanes I and J), but the relative intensity of the apolipoprotein B band was very low, and again the anti-apolipoprotein A1 antiserum only rarely precipitated detectable material (Table 1). The anti-LDL antibody also immunoprecipitated a 42 –44 ×103 molecule from the medium over cultured yolk sacks; this labelled molecule was not detected in the liver and it was only once found in the gut samples.
Products secreted by human yolk sack. SDS-PAGE slab gel electrophoretic analysis of [35S]methionine-labelled culture medium from a 51-day postfertilization embryo. Lane A: total secreted proteins; Lane B: anti-transferrin immunoprecipitation; Lane C: anti-AFP immunoprecipitation; Lane D: anti-LDL immunoprecipitation. In lane D, the major apolipoproteins are identified with reference to their apparent relative molecular mass (Mr). Some labelled material with the apparent relative molecular mass of AFP or albumin has also been precipitated by this antibody. The relative molecular mass markers are myosin: 200 ×103; phosphorylase b: 92 ·5 ×103; bovine serum albumin: 69 ×103; ovalbumin: 46 ×103; and carbonic anhydrase: 30 ×103.
Products secreted by human yolk sack. SDS-PAGE slab gel electrophoretic analysis of [35S]methionine-labelled culture medium from a 51-day postfertilization embryo. Lane A: total secreted proteins; Lane B: anti-transferrin immunoprecipitation; Lane C: anti-AFP immunoprecipitation; Lane D: anti-LDL immunoprecipitation. In lane D, the major apolipoproteins are identified with reference to their apparent relative molecular mass (Mr). Some labelled material with the apparent relative molecular mass of AFP or albumin has also been precipitated by this antibody. The relative molecular mass markers are myosin: 200 ×103; phosphorylase b: 92 ·5 ×103; bovine serum albumin: 69 ×103; ovalbumin: 46 ×103; and carbonic anhydrase: 30 ×103.
Secreted products of foetal liver; specimen at 61 days postfertilization. Secreted products of foetal gut; specimen at 63 days postfertilization. Lanes A and F are total secreted products. Lanes B and H are anti-AFP immunoprecipitates. Lane C and G are anti-transferrin immunoprecipitates, although in lane G no transferrin is detectable in this case. Lanes D and I are anti-apolipoprotein Al immunoprecipitates. Lanes E and J are anti-LDL immunoprecipitates. The major apolipoproteins are identified by their apparent molecular weight. Relative molecular mass standards as before, with the addition of lysozyme at 14 ·3 ×103.
Secreted products of foetal liver; specimen at 61 days postfertilization. Secreted products of foetal gut; specimen at 63 days postfertilization. Lanes A and F are total secreted products. Lanes B and H are anti-AFP immunoprecipitates. Lane C and G are anti-transferrin immunoprecipitates, although in lane G no transferrin is detectable in this case. Lanes D and I are anti-apolipoprotein Al immunoprecipitates. Lanes E and J are anti-LDL immunoprecipitates. The major apolipoproteins are identified by their apparent molecular weight. Relative molecular mass standards as before, with the addition of lysozyme at 14 ·3 ×103.
As expected from the work of previous authors, the synthesis of alphafoetoprotein and transferrin were, with one exception, confined to foetal tissues which contained endoderm cells (Table 1). In addition, we used [3H]leucine-labelled samples to extend these observations on foetal liver and the yolk sack: we specifically immunoprecipitated AFP, apolipoprotein Al, transferrin (70, 74, and 75 days postfertilization yolk sacks, one liver of unknown age), albumin (74 days postfertilization yolk sack, one liver of unknown age), and pre-albumin (same specimens). We also showed that two of these yolk sacks and this liver sample synthesized and secreted the 230 ×103 form of fibronectin (results not shown).
We attempted to identify the cell type which synthesized apolipoproteins, transferrin, albumin, and AFP in the secondary human yolk sack by studying the distribution of these molecules on tissue sections (see Materials and Methods). The anti-AFP antibody reaction was concentrated in the cytoplasm of the endoderm cells in two samples at 50 and 51 days postfertilization, confirming the results of previous authors (results not shown: see Albrechtsen et al. 1980). In contrast, the apolipoprotein Al and anti-transferrin antibodies reacted with the cytoplasm of both the endoderm cells and the outer epithelial cells, and they also reacted with the periphery of the nucleated erythrocytes (samples at 42,48, and 66 days postfertilization; Figs 7 –10). These three samples had been cultured for 19h in the absence of these serum products, in an attempt to elute and dilute out molecules which might have absorbed to the cell surface or which might have been taken up into the cytoplasm of these cells. This treatment reduced the already weak staining over mesenchyme and the lining of blood vessels, but it did not affect the relative staining intensity of the endoderm, the outer epithelial cells, and the erythrocytes. The weak albumin staining showed the same distribution on yolk sack samples taken at 50, 51, and 54 days postfertilization, and cultured for 20 h in the medium without albumin before processing. Erythrocytes from an embryo of a similar age have previously been observed to react with anti-albumin antibodies (Jacobsen et al. 1981).
Localization of apolipoprotein Al (Fig. 7), and transferrin (Fig. 9) on a transverse section of a 42-day postfertilization secondary yolk sack. This sack had been cultured for 19 h in the absence of these serum proteins (see Materials and Methods), e, endoderm; m, mesenchyme; ep, epithelium; b, blood cells. Scale bar, 50 μm.
Fig. 7. Reacted with anti-apolipoprotein Al antibody (1/50), and 1/100 sheep peroxidase conjugated second layer. The endoderm, blood, and epithelial layer are strongly stained.
Fig. 8. Two sections away and stained under identical conditions with the first antibody absorbed with apolipoprotein A1. Photographic procedures identical for both prints.
Fig. 9. Reacted with anti-transferrin antibody at 1/200, and peroxidase conjugated second layer at 1/100. The endoderm column and epithelial layer are strongly stained.
Fig. 10. Two sections away and stained under identical conditions with the first antibody absorbed with transferrin. Photographic procedures kept constant for both prints.
Localization of apolipoprotein Al (Fig. 7), and transferrin (Fig. 9) on a transverse section of a 42-day postfertilization secondary yolk sack. This sack had been cultured for 19 h in the absence of these serum proteins (see Materials and Methods), e, endoderm; m, mesenchyme; ep, epithelium; b, blood cells. Scale bar, 50 μm.
Fig. 7. Reacted with anti-apolipoprotein Al antibody (1/50), and 1/100 sheep peroxidase conjugated second layer. The endoderm, blood, and epithelial layer are strongly stained.
Fig. 8. Two sections away and stained under identical conditions with the first antibody absorbed with apolipoprotein A1. Photographic procedures identical for both prints.
Fig. 9. Reacted with anti-transferrin antibody at 1/200, and peroxidase conjugated second layer at 1/100. The endoderm column and epithelial layer are strongly stained.
Fig. 10. Two sections away and stained under identical conditions with the first antibody absorbed with transferrin. Photographic procedures kept constant for both prints.
We wished to exclude the possibility that it was the blood cells in these foetal organ samples which were synthesizing these apolipoproteins; foetal blood has been previously reported to synthesize β-lipoproteins which might contain these apolipoprotein species (Gitlin & Biasucci, 1969). The umbilical cord blood did not incorporate detectable radioactivity into either apolipoproteins Al or B, as judged by the failure to immunoprecipitate labelled molecules with the correct mobility with any of the anti-apolipoprotein antibodies. In the same experiment, and on the same gel, these molecules were immunoprecipitated from the culture medium over a yolk sack, when the yolk sack sample contained similar quantities of TCA-precipitable radioactivity.
In addition, one yolk sack stalk was fixed at 54 days postfertilization, and the endoderm tube was found to react with anti-AFP, anti-transferrin, and antialbumin antibodies.
DISCUSSION
1. Identity of the apolipoproteins
The anti-LDL antibody precipitated labelled molecules with the mobilities of apolipoproteins Al, and B from almost all secondary yolk sack samples (Table 1). The identity of synthesized human apolipoproteins Al and B was further supported by their precipitation with commercial antisera raised to these individual apolipoprotein species. The coprecipitation of several apolipoprotein species by both the anti-human LDL and anti-apolipoprotein A1 antisera does not necessarily indicate that all these apolipoproteins are bound together in a single particle, for at least the anti-LDL antiserum is known to recognize several different mouse apolipoproteins individually (immunoblotting: Shi & Heath, 1984). Coprecipitation can then be also due to specific recognition of determinants on a variety of apolipoproteins. Although it is already known that sialo groups can alter the mobility of the apolipoprotein E secreted by the foetal liver (Zannis et al. 1982); such changes have not been observed with apolipoproteins Al and B, and we consider that the mobilities of these two apolipoproteins are sufficiently distinct from that of other apolipoproteins to establish their identity in this study. It should be noted that we have not distinguished the various forms of apolipoprotein B in this study (e.g. Milne et al. 1984).
We can not identify the 42 –44 ×102 3 molecule, which is precipitated by the anti-LDL antibody; it is slightly smaller than apolipoprotein AIV (46 × 103, Bieseigel & Utermann, 1979), and slightly larger than the 40 ×103 form of apolipoprotein E which is secreted from foetal liver (Zannis et al. 1982).
2. The yolk sack as an early source of apolipoproteins
This study establishes that the human secondary yolk sack is a rich source of secreted apolipoproteins during the 5 –11th weeks of postfertilization development; over this time, the liver and the gut also synthesize and secrete apolipoproteins. There are several reasons for thinking that the yolk sack may be the sole source of these lipid carrier molecules during the second and third week of human development. First, the visceral yolk sack is the earliest organ to synthesize apolipoproteins in mouse development (Shi & Heath, 1984); the synthesis of apolipoproteins is detected on the 10th day of development, two days after this mouse tissue starts to synthesize alphafoetoprotein and transferrin (Adamson, 1982; Dziadek & Adamson, 1978; Dziadek & Andrews, 1983). Second, the secondary human yolk sack which we have studied appears to be the direct cellular descendant of the primary yolk sack (Luckett, 1978), and the primary human yolk sack might therefore synthesize apolipoproteins at the earliest stages of postimplantation development. Third, the liver and the gut of human embryos are not coherent organs during these two weeks (Moore, 1982). In humans,-the lining of the gut and the allantois is continuous with that of the yolk sack, and it is already known that these epithelia are distinct in the third week of postfertilization development; it is only inside the human yolk sack endoderm cell cytoplasm that embryonic serum proteins are found at this time (20-day postfertilization embryo, Jacobsen et al. 1981). The capacity of the gut lining to secrete serum products must therefore develop later. Fourth, it is a possibility that maternal apolipoproteins are unable to cross the placenta (Winkel, Snyder, MacDonald & Simpson, 1980). If these speculations are correct, then an intrinsic source of apolipoproteins from the human yolk sack may be essential to sustain the rapid cell divisions of early embryogenesis (e.g. mouse: Solter & Skreb, 1971; Snow, 1977).
In the mouse, it has been established that it is the endoderm layer of the yolk sack which is the source of apolipoproteins (Shi & Heath, 1984; Meehan et al. 1984). We have failed to establish that this is the case in the human, both because knowledge of a molecules distribution does not provide information about its site of synthesis, and because we considered it impractical to cleanly separate the convoluted endoderm from the other yolk sack tissues. It should be noted that apolipoprotein synthesis is not always restricted to endoderm cells, and that at least apolipoprotein E can be synthesized by human blood monocytes, and the mRNA for this apolipoprotein is present in human adult leucocytes and kidney (Basu et al. 1982; Wallis et al. 1983). It therefore remains a possibility that the intensive localization of apolipoproteins in both the endoderm, the outer epithelial layer, and the nucleated erythrocytes of the human yolk sack does reflect synthesis of apolipoproteins by all these cell types. However, our inability to detect the synthesis of apolipoproteins A1 and B by foetal blood cells makes it unlikely that blood cells are a source of these apolipoproteins; it remains a possibility that the haemopoietic stem cells in the yolk sack and liver synthesize these proteins and that such cells never enter the foetal circulation.
In general, our observations on the synthesis of the other foetal serum proteins is consistent with previous authors (Table 1, see Gitlin & Biasucci, 1969; Gitlin & Perricelli, 1970; Gitlin et al. 1972; Nishi, 1970). On one occasion, labelled AFP was detected in the culture medium over trophoblast; a similar case of exceptional AFP synthesis has been observed before in 1/14 placental samples (Gitlin et al. 1972). One stomach sample also secreted AFP; we do not regard this as a reliable result because we can not exclude the possibility that small fragments of the neighbouring foetal liver contaminated this specimen.
There was variation in our ability to detect the synthesis and secretion of both apolipoproteins and the other serum proteins from gut samples; a sample usually secreted all or secreted none. This variability could not be explained either by the size of the fragments or their age (Table 1). It is likely that there are local variations in the secretion of these molecules along the length of the foetal gut; for in the adult gut apolipoprotein synthesis is inducible (Bisgaier & Glickman, 1983), and there may be local distinctions along the length of the gut (Green et al. 1982). The disruption caused by the vacuum abortion procedure made it impossible to routinely recognize the region of the gut which was incubated with label.
This study extends the list of products which are secreted in common by the mouse visceral yolk sack and the human secondary yolk sack. These two organs have a very different anatomical relations with the embryo and the other extraembryonic membranes in the two species, and it is still unclear how either iron or lipid reaches these organs, to be picked up by the transferrin and apolipoprotein carrier molecules which they synthesize and secrete.
CONCLUSION
During the 5 –11th weeks of postfertilization human development, the secondary yolk sack is a source of apolipoproteins Al and B.
These lipid carrier molecules are also synthesized by the liver and the gut over this period of embryonic development.
ACKNOWLEDGEMENTS
We thank the Blood Transfusion Unit of the John Radcliffe Hospital for fresh venous blood. The Cancer Research Campaign generously supported these studies. Dr W-K. Shi was on sabbatical leave from The Shanghai Institute of Cell Biology, Academia Sinica, China, supported by the Chinese Academy of Sciences: Royal Society exchange scheme, the Henry Lester Fund, and the British Universities China Committee.