Ex vivo analysis of acinar and endocrine cell development in the human embryonic pancreas



In contrast to the considerable body of data on pancreas development in rodents, information on pancreas development in humans is scant. We previously described a model in which mature beta cells developed from human embryonic pancreas: human embryonic pancreas was grafted under the kidney capsule of scid mice, beta cells were then seen to develop in the graft. Here, we showed that not only beta cells, but also other endocrine cells, acinar cells and ducts develop in this model. We then used this model to probe the mechanisms underlying acinar and beta cell development in the human embryonic pancreas. BrdU pulse/chase experiments produced evidence of clonal acinar cell development: the first acinar cells to appear proliferated, thereby expanding the acinar cell population. In contrast, beta cell development was regulated by the proliferation of pancreatic progenitor cells, followed by beta-cell differentiation. We then showed that early progenitors expressing PDX1 proliferated, whereas late endocrine progenitors expressing Ngn3 did not. This proliferative capacity of early endocrine progenitor cells in embryonic human pancreas may hold promise for obtaining human beta-cell expansion. Developmental Dynamics 234:339–345, 2005. © 2005 Wiley-Liss, Inc.


The mature pancreas contains two types of tissue: endocrine islets composed of cells that produce hormones such as insulin (beta cells), glucagon (alpha cells), somatostatin (delta cells), and pancreatic polypeptide (PP cells); and exocrine tissue composed of acinar cells that produce enzymes (e.g., carboxypeptidase-A) secreted via pancreatic ducts into the intestine. The pancreas develops from buds derived from the digestive tract epithelium. The buds increase in size, and the epithelial cells differentiate into endocrine and exocrine cells. In recent years, considerable progress has been made in understanding the basic processes that control pancreas development in rodents. Studies of genetically modified mice have established a hierarchy of transcription factors regulating pancreas organogenesis and have produced information on endocrine and acinar cell differentiation (Wilson et al., 2003). Signals of mesodermal origin have been shown to play a major role in controlling the proliferation of immature pancreatic epithelial cells and their cytodifferentiation to exocrine or endocrine tissue (Scharfmann, 2000; Kim and Hebrok, 2001; Lammert et al., 2003). Finally, recent data clarified the importance of beta cell proliferation during postnatal life in rodents (Dor et al., 2004).

Whereas the control of pancreas development in rodents is gradually yielding its secrets, the embryonic development of the human pancreas remains largely unknown. Obstacles to human studies include limited availability of human tissues, particularly from the first trimester of prenatal life. In addition, the vast majority of data came from static histological studies (Stefan et al., 1983; Bouwens et al., 1997; Bouwens and Pipeleers, 1998; Polak et al., 2000). The scarcity of data on human pancreas development can be ascribed to the paucity of experimental models. Recently, we designed an ex vivo model in which fully mature endocrine cells can develop from human embryonic pancreas. Briefly, when human embryonic pancreases (6–9 weeks of development), which contain very few beta cells, are engrafted into NOD/scid mice, the human pancreatic tissue grows rapidly, its weight increasing 200-fold within 6 months, and beta cells differentiate, leading to a 5,000-fold increase in the beta-cell count. Finally, the human beta cells are mature, being capable of regulating the blood glucose levels of mice deficient in endogenous beta cells (Castaing et al., 2001).

In the present study, we used the above-described model to investigate the development of other human pancreatic cell types. We found that human acinar cells developed by clonal proliferation of differentiated cells. With beta cells, the sequence was reversed: progenitors underwent proliferation followed by differentiation.


Differentiation of All Pancreatic Cell Types Upon Grafting of Immature Human Embryonic Pancreas to Scid Mice

We have shown previously that immature human embryonic pancreas grafted under the kidney capsule of scid mice undergoes growth, with development of beta cells (Castaing et al., 2001). Here, we first determined whether all major pancreatic cell types (endocrine, acinar, and ductal cells) developed in human embryonic pancreas grafted into scid mice. Before transplantation, very few cells expressing insulin, glucagon, or somatostatin were detected in the human embryonic pancreases (Fig. 1A and C). At that stage, and as previously shown (De Krijger et al., 1992; Polak et al., 2000), the few insulin-containing cells frequently coexpressed glucagon (Fig. 1A, arrows). In pancreatic tissue examined 21 weeks after transplantation, numerous endocrine cells had developed and were associated in islet-like structures containing insulin-, glucagon-, and somatostatin-positive cells (Fig. 1B and D). We next examined exocrine development. Before transplantation, the acinar tissue was poorly developed. Cells expressing trypsin were not detected, whereas a few carboxypeptidase-A-positive cells were found (Fig. 1E,G). When human embryonic pancreases were grafted into scid mice, the acinar tissue developed rapidly: after 1.5 months, cells expressing trypsin and carboxypeptidase-A were seen to form acinar structures (Fig. 1F,H). We also performed co-immunostaining for carboxypeptidase-A and insulin on human embryonic pancreases that developed in mice during 4, 8, 12, and 21 weeks. As shown in Figure 2, we never found cells positive for both carboxypeptidase-A and insulin. This result indicates that in this system, acinar/endocrine intermediate cells are rare or absent. Before transplantation, duct-like structures staining for cytokeratin were also found. However, these cells stained negative for both CFTR and CA19-9, two markers for mature duct cells (Fig. 1I,K). In contrast, when human embryonic pancreases were grafted into scid mice, duct-like structures differentiated into mature duct cells expressing both CFTR and CA19-9 (Fig. 1J,L).

Figure 1.

Immunohistochemical study of human embryonic pancreases transplanted into scid mice. Human embryonic pancreases were studied before grafting (A, C, E, G, I, K) and after grafting under the kidney capsule of scid mice for 21 weeks (right panel B, D, J, L) or for 6 weeks (right panel F, H). Endocrine development was evaluated after anti-insulin (red) (A, B, C, D), anti-glucagon (green) (A, B), or anti-somatostatin (green) (C, D) staining. Development of acinar tissue was studied using anti-trypsin (green) (E, F) and anti-carboxypeptidase-A (green) (G, H) antibodies. Ductal cell development was investigated using anti-cytokeratin (green) (I, J, K, L), anti-CFTR (red) (I, J), and anti-CA19.9 (red) (K, L) antibodies. In A and C, the nuclei were stained with Hoescht. In A, the arrow points to a cell positive for both insulin and glucagon. In C, the arrowhead shows a cell positive for insulin. Scale bar = 25 μm.

Figure 2.

Relationship between endocrine and acinar tissue during transplantation of human embryonic pancreases. Human embryonic pancreases were grafted to scid mice and analyzed 4 (A, B), 8 (C, D), 12 (E, F), and 21 (G, H) weeks after transplantation. Anti-insulin (red) and anti-carboxypeptidase-A (green) antibodies were used. In A, C, E, and G, the green staining corresponding to carboxypeptidase-A is shown. In B, D, F, and H, double staining corresponding to carboxypeptidase-A (in green) and insulin (in red) is shown. In A, C, E, and G, the areas corresponding to insulin-positive cells are circled. Scale bar = 25 μm.

Blood vessels represent another important component of the mature pancreas. We studied the type of vessels present in human embryonic pancreas grafted into scid mice. Before grafting, the human embryonic pancreas contained vessels located in the mesenchymal tissue that stained positive for CD34 (Fig. 3A). Two months after transplantation, the surface of the graft was covered by prominent blood vessels from the renal capsule (Fig. 3B). We next sectioned the grafts to determine the nature of the vessels present within the graft, using an anti-CD34 antibody that labels human but not mouse endothelial cells (Fig. 3C). As shown in Figure 3D, the vast majority of endothelial cells present in human embryonic pancreas grafted into scid mice were recognized by anti-CD34 antibody, indicating that these cells were of human origin.

Figure 3.

Nature of the vessels in transplanted human embryonic pancreases. Human embryonic pancreases were studied before (A) and 8 or 12 weeks (B and D, respectively) after transplantation. Before transplantation, human endothelial cells were located in the mesenchymal tissue. After transplantation, blood vessels from the renal capsule were visible at the graft surface (B). Within the graft, the endothelial cells were of human origin, as shown by their labeling with anti-CD34 antibody, which recognizes human endothelial cells (D) but not endothelial cells present in the mouse pancreas (C). In B and C, the arrows point to vessels. Scale bar = 25 μm (A, C, D), 1 mm (B).

PDX1 Expression in Human Embryonic Pancreas Grafted Into scid Mice

In the rodent pancreas, during the first steps of pancreas development, all pancreatic epithelial cells express the transcription factor PDX1. Subsequently, PDX1 expression is restricted to beta cells; the other endocrine, acinar, or duct cells do not express PDX1 (Ohlsson et al., 1993; Offield et al., 1996). In the present study, before transplantation, PDX1 was expressed by pancreatic epithelial cells that stained positive for cytokeratin (Fig. 4A and B). Human embryonic pancreases were grafted into scid mice and examined 4 or 21 weeks later. Four weeks after grafting, PDX1 was found in the earliest insulin-positive cells (Fig. 4C) and remained present in pancreatic epithelial cells that stained positive for cytokeratin (Fig. 4D). After 21 weeks, PDX1 was detected in the nuclei of the beta cells, which were clustered into islet-like structures (Fig. 4E). At that stage, epithelial cells that formed duct-like structures stained positive for PDX1. However, the signal was weaker in duct-like cells than in beta cells (Fig. 4F). This pattern of expression that developed over 21 weeks in beta cells and duct cells of human embryonic pancreases in scid mice resembled the pattern of PDX1 expression found in human adult pancreas (Fig. 4G,H).

Figure 4.

Expression pattern of PDX1 in human embryonic pancreas during transplantation. A–F: human pancreases at 8 weeks of development before transplantation (A, B) and 4 weeks (C, D) or 21 weeks (E, F) after transplantation; G, H: human adult pancreas. In A–H, PDX1 was revealed in green. In A, C, E, and G, insulin was revealed in red. In B, D, F, and H, cytokeratin was revealed in red. Scale bar = 25 μm.

Human Acinar Cells Develop by Clonal Proliferation of Differentiated Cells, Whereas Human Beta Cells Develop by Proliferation of Progenitors Followed by Differentiation

Our next objective was to determine how acinar and endocrine cells developed in our model. For that purpose, mice were grafted with human embryonic pancreases and next injected twice a day for 5 days with BrdU, which labels cells in the S-phase. The mice were killed 2 h, 17 days, or 53 days after the last BrdU injection. As shown in Figure 5A, at the end of the pulse period, a large number of human pancreatic cells had incorporated BrdU, showing that the labeling method was efficient. More specifically, 69 ± 14% (n = 3 grafts) of PDX1-expressing cells stained positive for BrdU at the end of the pulse. BrdU labeling was also seen in some of the acinar cells that were positive for carboxypeptidase-A and present at the end of the pulse period (Fig. 5B). After a 17-day chase period (Fig. 5C), the number of BrdU-positive cells was smaller, and the vast majority of carboxypeptidase-A-positive cells stained negative for BrdU (Fig. 5C). The decrease in the number of BrdU-positive cells was even more marked after a 53-day chase period, when all carboxypeptidase-A-positive cells were BrdU-negative. The finding that the acinar cells were positive for BrdU at the end of the pulse period and negative 17 and 53 days later indicates that acinar cells developed by division of carboxypeptidase-A-positive cells present at the end of the pulse period and not by differentiation of progenitors.

Figure 5.

Human acinar cells developed by proliferation of differentiated cells, whereas beta cells developed by proliferation of progenitor cells followed by differentiation. Human embryonic pancreases were transplanted into scid mice. Three days later, BrdU was injected twice a day for 5 days. The grafts were examined 2 h (A, B, E), 17 days (C, F), or 53 days (D, G) after the last BrdU injection. The tissues were sectioned and stained for BrdU (red) (A-G); PDX1 (green) (A); carboxypeptidase-A (B–D) (green); or insulin (E–G) (green). In B, the arrows point to cells positive for both carboxypeptidase-A and BrdU. In F and G, the arrows point to cells positive for both insulin and BrdU. Scale bar = 25 μm (A), 12.5 μm (B–G)

We performed identical experiments for beta cells. As shown in Figure 5E, at the end of the pulse period, the few insulin-positive cells stained negative for BrdU. Seventeen days later, the number of insulin-positive cells had increased, whereas the BrdU signal had been diluted from the vast majority of the cells. However, at that point, we clearly detected insulin-positive cells that stained positive for BrdU (Fig. 5F). This was also the case after a 53-day chase period (Fig. 5G). The fact that insulin-positive cells were negative for BrdU at the end of the pulse period and positive 17 and 53 days later indicates that human beta cells developed by differentiation of progenitors that proliferated during the pulse period.

Human PDX1-Expressing Cells But Not Ngn3-Expressing Cells Proliferate in the Human Embryonic Pancreas

In rodents, beta cells derive from PDX1-expressing progenitor cells, which transiently express Ngn3 before they differentiate into beta cells (Gradwohl et al., 2000; Herrera, 2000; Gu et al., 2002). The above-reported data indicate that human beta cells derive from PDX1-expressing cells that first proliferate. We, therefore, investigated the Ngn3 status of the proliferating PDX1-positive progenitor cells. As shown in Figure 6A, when human embryonic pancreases were incubated for 2 h with BrdU, cells that stained positive for both PDX1 and BrdU were frequently detected, indicating a high proliferative potential of the PDX1-positive progenitors. We then investigated whether human Ngn3-expressing progenitor cells were undergoing proliferation. As shown in Figure 6B, Ngn3-positive cells were present in the human embryonic pancreases; they were located within the ductal tree and stained positive for PDX1 (data not shown). However, when human embryonic pancreases were pulsed for 2 h with BrdU, we could not detect Ngn3-expressing cells that stained positive for BrdU (Fig. 6B). To increase the chance of finding such double-positive cells, we investigated the proliferative potential of Ngn3-positive cells using an antibody to the nuclear antigen Ki67, which is expressed in late G1-, S-, G2, and M phases but not in the quiescent phase G0. As shown in Figure 6C, none of the Ngn3-expressing cells were positive for Ki67. Taken together, our data indicate that the early PDX1-expressing progenitor cells represented the main amplification pool in the beta-cell lineage in human embryos.

Figure 6.

In vitro, cells expressing PDX 1 proliferate, whereas cells expressing Ngn3 do not. Human pancreases at 8 weeks of development were dissected and cultured for 2 h with BrdU. The tissues were sectioned and stained for: A: PDX 1 (green) and BrdU (red). The arrows point to cells positive for both PDX1 and BrdU; B: Ngn3 mRNA detected after in situ hybridization (in blue) and BrdU (red). C: Ngn3 mRNA detected after in situ hybridization (in blue) and Ki67 (brown). Scale bar = 25 μm.


In this work, we further characterized a model in which immature human embryonic pancreas develops into mature pancreatic tissue. We then used this model to determine how human acinar and beta cells develop in the human embryonic pancreas.

In rodents, information is available on pancreatic cell development during prenatal and postnatal life (Pictet and Rutter, 1972; Tsubouchi et al., 1987; Finegood et al., 1995; Bhushan et al., 2001; Dor et al., 2004), whereas few data have been obtained in humans. We show here that, when human embryonic pancreas is grafted under the kidney capsule of scid mice, all pancreatic cell types develop. Furthermore, before grafting, nearly all pancreatic epithelial cells express the transcription factor PDX1, a marker of pancreatic progenitor cells. Once the tissue differentiates, PDX1 is excluded from acinar cells and from glucagon-expressing cells (data not shown); in contrast, all beta cells express PDX1. This pattern of expression is identical to that found in vivo in mice (Ohlsson et al., 1993). Interestingly, while PDX1 is excluded from mature duct cells in rodents (Ohlsson et al., 1993), it remains present in human duct cells that develop in the graft. This has been found also for duct cells present in the human adult pancreas (Heimberg et al., 2000; the present study). The difference in terms of PDX1 expression between human and rodent adult pancreatic duct cells is interesting. While PDX1 is known to be crucial both for progenitor cell development in the embryonic pancreas and for postnatal beta-cell function, in rodents and humans (Offield et al., 1996; Stoffers et al., 1997; Ahlgren et al., 1998; Hani et al., 1999), its specific function in human adult duct cells remains unknown. Defining specific targets for PDX1 in these cells should provide insight into the function of PDX1 in human adult duct cells.

Theoretically, several mechanisms could explain the expansion of the mature-cell population during development. One mechanism is clonal cell development, with differentiation of a cell to a specific type followed by clonal proliferation. Another is proliferation of specific progenitors followed by maturation, with loss of the ability to proliferate. There is firm evidence that in rodents, pancreatic progenitor cells expressing PDX1 proliferate at a fast rate between E9 and E14 (Bhushan et al., 2001). If the cells enter the acinar pathway, they continue to divide (Pictet and Rutter, 1972; Duvillie et al., 2003). In contrast, once cells enter the endocrine pathway and start to express Ngn3, their proliferation rate decreases sharply (Jensen et al., 2000). Toward the end of prenatal life and after birth, beta cells reenter the cell cycle: the current hypothesis is that, after birth, terminally differentiated beta cells remain capable of substantial proliferation, which is sufficient to ensure beta cell homeostasis during adulthood (Dor et al., 2004). Such data were obtained in rodents by immunohistological studies after BrdU pulses at specific stages of development (Jensen et al., 2000; Bhushan et al., 2001) or by genetic lineage tracing (Dor et al., 2004). Such methods have not been used to investigate human pancreatic acinar and endocrine cell formation, because no appropriate experimental system was available. Most of the data on the proliferative potential of specific human pancreatic cell types was obtained after double-staining of human pancreatic sections with a marker for a specific pancreatic cell type coupled to an antibody against the nuclear antigen Ki67, which is expressed in the late G1-, S-, G2-, and M- phases but not in the quiescent phase G0 (Gerdes et al., 1984). With this method, a small percentage of cells positive for both insulin and Ki67 were found during prenatal life. In contrast, no cells positive for insulin and Ki67 were detected in human adult pancreas (Bouwens and Pipeleers, 1998). Here, we used BrdU labeling to determine whether human acinar and beta cells developed by proliferation of differentiated cells or of progenitors. We first created conditions in which the vast majority of PDX1-positive progenitor cells in human embryonic pancreases were labeled with BrdU. Under these conditions, some acinar cells were labeled, whereas the few insulin-positive cells present during the labeling period did not incorporate BrdU. Next, we chased BrdU for 17 or 53 days. If a specific cell type develops by proliferation of preexisting mature cells, absence of BrdU in this specific cell type is expected at the end of the chase period. This pattern was observed in our study for acinar cells. In contrast, if a specific cell type develops by proliferation of progenitor cells during the pulse period followed by differentiation, presence of BrdU-positive mature cells is expected at the end of the chase period. This was the case for beta cells in our study. Taken together, our data indicate that, during embryonic life, pancreatic acinar cells develop by proliferation of preexisting acinar cells, whereas beta cells develop by proliferation of progenitor cells followed by differentiation to the beta-cell phenotype.

In rodents, beta cells develop from PDX1-expressing progenitor cells that transiently express Ngn3 before differentiating into endocrine cells (Gradwohl et al., 2000; Herrera, 2000; Gu et al., 2002). Therefore, it was important to determine the type of endocrine progenitors that proliferate during embryonic life in human. We found that PDX1-expressing progenitor cells incorporated BrdU, whereas Ngn3-expressing cells neither incorporated BrdU nor expressed Ki67. It has been suggested that Ngn3-positive cells in rodents may consistently originate from Ngn3-negative cells during development, not from proliferation of preexisting Ngn3-expressing cells (Gu et al., 2002). Our present data, indicating that in humans, Ngn3-positive cells neither incorporate BrdU nor express Ki67, further support this suggestion.

Taken together, our data indicate that PDX1-expressing progenitor cells proliferate in human embryos. These cells, which may be used as a source of beta cells, will be further studied in the future. Once they develop into acinar cells, they continue to proliferate. In contrast, if they develop into endocrine cells, they stop proliferating before expressing Ngn3. Defining the factors that activate the proliferation of PDX1-expressing progenitor cells might lead to new strategies for cell therapy in diabetes.


Human Tissues

Human pancreases were extracted from embryonic tissue fragments obtained immediately after elective termination of pregnancy performed by aspiration between 7.5 and 9.5 weeks of development, in compliance with French legislation and the guidelines of our institution. Warm ischemia lasted less than 30 min. Gestational ages were determined on the basis of time since the last menstrual period, crown-rump length measured by ultrasonography, and hand and foot morphology. Six pancreases were dissected, immediately fixed, and embedded in paraffin. Fourteen pancreases were grafted into mice with severe combined immunodeficiency (scid), as described below.

Scid Mice and Transplantation Procedure

Scid mice (Charles River Laboratories, L'arbresles, France) were bred in isolators supplied with sterile, filtered, temperature-controlled air. Cages, bedding, and drinking water were autoclaved. Food was sterilised by X-ray exposure. All manipulations were performed under a laminar flow hood. Embryonic pancreases (7.5–9.5 weeks of development) were implanted under the left kidney capsule of 6- to 8-week-old scid mice, using a dissecting microscope, as previously described (Castaing et al., 2001). At different time-points after transplantation (ranging from 8 days to 5 months), mice were killed and the grafts were removed, fixed in 10% formalin, and embedded in paraffin. Similar exocrine or endocrine development was observed when different pancreatic stages (between 7.5 to 9.5 weeks of development) were used for the graft.


Sections 4 μm in thickness were cut on gelatinized glass slides, deparaffinized in toluene, and rehydrated. For immunofluorescence staining, sections were microwaved in citrate buffer 10 mM, pH 6, and permeabilized for 20 min in TRIS-Buffered Saline (TBS) containing 0.1% Triton. Non-specific sites were blocked for 30 min in TBS containing 3% BSA and 0.1% Tween 20, and sections were incubated overnight at 4°C with primary antibodies. The sections were then washed and incubated for 1 h at room temperature with the appropriate secondary antibodies, labelled with two different fluorochromes. The primary antibodies were mouse anti-human insulin (Sigma Aldrich, 1/1,000), rabbit anti-glucagon (DiaSorin, Antony, France,1/1,000), rabbit anti-human somatostatin (Dako, Trappes, France,1/1,000), rabbit anti-human trypsin (a gift from Dr. Figarella, Marseille, France, 1/500), rabbit anti-bovine carboxypeptidase-A (AbCys, Paris, France, 1/600), mouse anti-human CFTR (Genzyme, Cambridge MA, 1/20), mouse anti-human CA 19.9 (Dako, 1/50), mouse anti-BrdU (Amersham, Saclay, France, 1/2), mouse anti-pancytokeratin (Sigma, 1/100), mouse anti-Ki67 (Immunotech, Roissy, France, 1/50), mouse anti-human CD 34 antibody (Beckman Coulter, Roissy, France), rabbit anti-bovine keratin (Dako, 1/500), mouse anti-porcine vimentin (Dako, 1/50), and rabbit anti-human PDX1. The rabbit anti-PDX1 antibody was raised against the synthetic peptide (C) SPQPSSVAPRRPQEPR conjugated to KLH.

Fluorescent secondary antibodies were fluorescein anti-rabbit and anti-mouse (Immunotech 1/200) and Texas-red anti-rabbit and anti-mouse (Immunotech 1/200). For CD34 and Ki67 staining, the streptavidin-biotin-peroxidase complex was used with diaminobenzidine as the chromogen, which produced brown staining.

BrdU Labeling and In Vivo Chase

Scid mice were grafted with human embryonic pancreases. Twice daily injections of BrdU (100 μg/g body weight) were started 3 days later and continued for 5 days. The mice were killed 2 h, 17 days, or 53 days after the last injection. The grafts were dissected, fixed in formalin, embedded in paraffin, and analyzed by immunohistochemistry.

Short-Term Culture

Human embryonic pancreases were cultured on Millipore filter inserts in 12-mm tissue-culture plates in 2 ml RPMI 1640 medium (Gibco, Cergy Pontoise, France) containing penicillin (100 U/ml), streptomycin (100 μg/ml), and 10% heat-inactivated calf serum, for 2 h at 37°C in a humidified atmosphere of 95% air/5% CO2. BrdU (3 μg/ml) was present in the medium during the 2-h culture period.

In Situ Hybridization

Tissues were fixed at 4°C in 4% paraformaldehyde in PBS, cryoprotected in 15% sucrose-PBS at 4°C overnight, embedded in 15% sucrose-7.5% gelatin in PBS, and frozen at −50°C in isopentane. Cryosections 14 μm in thickness were prepared. In situ hybridization was done as previously described (Herzog et al., 2001) using a human Ngn3 probe (a gift from Dr. Ravassard, Paris), and colorimetric revelation was performed with 5-bromo-4-chloro-3-indolyl phosphate (Roche Diagnostic, Meylan, France) and nitroblue tetrazolium (Roche Diagnostic) to obtain a blue precipitate. In situ hybridization was followed by immunohistochemistry using mouse anti-BrdU or mouse anti-Ki67 antibodies (Immunotech).


We are indebted to the medical staff in the Department of Gynecological surgery at the Hospital, R. Debé (Paris), A. Béclère (Clamast), and L. Mourier (Colombes) for providing human embryonic tissues. We are deeply grateful to Virginie Aiello for technical assistance.