Main

Because it is needed for the synthesis of deoxythymidylate(1), Cbl plays a key role in cell growth and is, therefore, essential for embryo-fetal development. The transport of Cbl involves three binding proteins, HC, IF, and TC II. The endocytosis of blood Cbl is mediated by a TC II receptor. A receptor for IF is also involved in the transcytosis of Cbl in the distal ileum of adults. IF and IF-Cbl receptor are also detected in fetal gastric and intestinal mucosa, respectively(2,3). In addition, the IF-Cbl receptor is present all along the intestine and in the colon before 15 wk of gestational age(2).

The delivery of Cbl from maternal blood to the fetus is still not fully understood. Little is known about the homeostasis of Cbl in early pregnancy and of the respective role of the Cbl-binding proteins. In congenital TC II deficiency, fetal growth is normal as long as a transplacental passage of maternal TC II occurs(47). A receptor for TC II has been described and purified from human, rabbit, and rat placental membranes. A receptor for IF-Cbl has also been detected in the rat visceral yolk sac and placental membranes(5,8). The expression of this IF-Cbl receptor declined dramatically from early to late gestation in the visceral yolk sac, whereas its activity rose in the placental membranes(5,8).

The amniotic fluid contains the three binding proteins of Cbl(9), and its Cbl concentration is of the same order of magnitude as maternal blood(7) and may be reduced in the case of neural tube defects(10). In addition, we have previously shown that the IF concentration in amniotic fluid decreases with gestational age. Because amniotic fluid is swallowed by the fetus, it may be assumed that Cbl is assimilated by the fetal intestine and that IF of amniotic fluid participates in Cbl delivery to the embryo-fetus.

To assess this issue, it remains to be determined that IF is secreted by the fetal stomach and that its binding to Cbl leads to an IF-Cbl receptor-mediated absorption of Cbl by the fetus.

The present work evaluated this hypothesis by studying the IF secretion and the Cbl absorption, respectively, in human fetal stomachs and intestines xenografted into nude mice.

METHODS

Surgical procedures

Animals. Pathogen-free congenitally athymic nude mice, Swiss nu/nu (IFFA CREDO Breeding Laboratories, L'Arbresle, France), aged 6 to 10 wk and weighing 20 to 25 g, were used in these experiments. All surgery was performed under general anesthesia with intraperitoneal ketamine (0.1 mg/g body weight).

Xenografting. Normal stomachs and intestines were obtained after voluntary induced abortion, carried out by the aspiration method, between the eighth and 10th week of amenorrhea. This protocol was submitted and accepted by the National Ethical Committee(6). The organs, selected under a Zeiss surgical microscope and kept in an isotonic solution at 4°C until grafting, were microsurgically implanted into pouches made within the mouse's lateroinferior abdominal walls. The posterior wall of the pouches was formed by the inferior epigastric vessels and the peritoneum and its anterior wall by the abdominal wall muscle layers. The grafts were sewn into place, in close contact with the epigastric vessels, with three fine stitches.

Assimilation of [57Co]Cbl by the intestinal xenografts (Schilling test). Eight weeks after grafting, 10 mice were given an intramuscular injection of 20 µL of unlabeled OH-Cbl (90 nmol/mL). Under laparotomy, tracer solutions (0.25 mL) were then instilled into the intestinal transplant lumen. The following two types of tracer solutions were used:

  1. 1

    an IF-[57Co]Cbl solution prepared by incubating 200 µL of gastric juice collected from the embryonic stomach No. 4 with a mixture of 1.5 pmol of [57Co]Cbl (450 000 cpm) and 300 pmol of cyanocobinamide for 20 min at room temperature (RT);

  2. 2

    a [57Co]Cbl solution consisting of 1.5 pmol of [57Co]Cbl in 0.25 mL distilled water.

Urines were collected for 24 h into miniature collectors tied onto the mouse's back and connected to small catheters passed through the membranous urethrae into the bladders. To avoid any photosensitive conversion of the labeled Cbl, urine collection was carried out in the dark.

Collection of the gastric and intestinal juices for biochemical studies. Six to 8 wk after grafting, collection of the gastric (six mice) and small intestinal (two mice) juices was performed by aspirating these digestive fluids under laparotomy. The gastric juice pH were then measured, and the samples of gastric and intestinal fluids were stored at -20°C in 0.2 mL aliquots.

Morphological methods

From each of the human gastric and intestine grafts removed, samples were fixed in ethanol/acetic acid (3/1, vol/vol) for 16 h and then routinely processed and embedded in paraffin. Five-micrometer-thick sections of all the paraffin blocks were stained with hematoxylin-eosin. Immunohistochemical localization of IF was carried out with a rabbit antiserum to human IF (diluted 1/200) by use of a sensitive three-step indirect immunophosphatase technique(11).

To assess the respective contribution of the host and donor cells to the transplant structure, detection of cell genotypes was performed as previously described(12) by in situ hybridization with biotinylated probes synthesized by the random primer labeling method using sonicated total genomic mouse or human DNA as templates.

Preparation of the probes. Total human or mouse genomic DNA (Clontech, France) was fragmented into 200-1000 bp-long pieces by sonication. Four hundred nanograms of DNA in 22 µL of distilled water mixed with 4 µL of a 90-U OD/mL solution of oligonucleotide hexamers (Pharmacia, France) in 0.01 M Tris-HCl was denatured for 10 min in a boiling water bath and then rapidly cooled on ice. Synthesis of the probes was carried out for 16 h at RT using 4 U of the Klenow fragment of Escherichia coli DNA polymerase (Pharmacia, France); 10 µL of a 0.3-mM dATP, dCTP, dGTP solution (Pharmacia, France) in 0.250 M Tris-HCl, pH 8.0; 0.025 M MgCl2; 3 µL of a 0.4-mM Bio-11-dUTP solution (Sigma Chemical Co., France) in the same buffer; and 10 µL of a 1-M HEPES-NaOH, pH 6.6. The reaction was stopped by adding 2 µL 0.5 M EDTA.

Biochemical methods

The protein content of gastric and intestinal fluids as well as that of one mucosal homogenate was determined by the method of Bradford with BSA as standard(13).

The mucosal homogenate extract was obtained by Potter homogenization of biopsy specimens in the presence of 50 µL of 20 mM Tris-HCl, pH 7.4, containing 0.15 M NaCl, 0.01% (wt/vol) sodium azide, 0.4 mM phenylmethylsulphonyl fluoride, and 0.5% Triton X-100 as previously described(14). After an overnight incubation period, the sample was centrifuged for 60 min at 14 000 × g and the binding capacity of the supernatant to Cbl-IF was assessed.

Gel filtration of Cbl-binding proteins of gastric fluids. Twenty to 50 µL of each gastric juice was incubated successively with 74 pmol of cyanocobinamide and 0.37 pmol of [57Co]-labeled Cbl for 5 and 20 min, respectively, in a total volume of 1 mL in 20 mM Tris-HCl buffer, pH 7.4, containing 0.7 M NaCl and 0.01% (wt/vol) NaN3, under rotative agitation at RT. Under these conditions, only IF is labeled with [57Co]Cbl, because cobinamide saturates HC. The same protocol was repeated in the absence of cobinamide to study both IF and HC saturated with labeled Cbl. The samples were run in a Superose 12 column (Pharmacia Fine Chemicals, Uppsala, Sweden) measuring 1 × 30 cm, eluted with the Tris-HCl buffer described above at a flow rate of 0.5 mL/min and using a Gilson HPLC pump model 305. Gamma-ray counting of [57Co]Cbl was performed in 1-min collected fractions of the eluate. The samples were also run in a Sephacryl S300 column (Pharmacia, Uppsala, Sweden) whose dimensions were similar to those of the Superose column. To collect the radioactive Cbl-binding protein peaks for isoelectrofocusing analysis, 50-µL samples were injected. The Cbl-binding proteins were not saturated under these conditions. Elution was performed with the same buffer at a flow rate of 0.3 mL/min. Dextran blue and [57Co]Cbl were used to determine the void volume and the total volume of the column, respectively.

Isoelectrofocusing. The fractions corresponding to the peaks of protein-bound [57Co]Cbl eluted in gel filtration were pooled and studied by isoelectrofocusing in a 110-mL LKB column, type 8101 (LKB, Bromma, Sweden) packed with carrier ampholytes of pH ranging from 3.0 to 9.0, at a final concentration of 1% in a 0-50% (wt/vol) sucrose gradient containing 3 M urea(14).

IF-Cbl receptor activity in the intestinal mucosa. Receptor activity was also determined using gel filtration as previously described(14). In short, the mucosal homogenate (50 µL) was incubated with a [57Co]Cbl-IF fraction collected by Sephacryl S300 gel filtration of fetal gastric juice in the presence of either 1 mM CaCl2 or 10 mM EDTA for 1 h at RT. The sample was then injected into a Sephacryl S300 gel filtration column as described above.

Determination of labeled vitamers excreted in the urine of recipient mice. Cbl were extracted with hot ethanol to denature and to precipitate the corrinoid-binding proteins(15,16). To prevent the nonspecific binding of OH-Cbl to histidine residues of proteins, the samples were preincubated for 2 h at RT with an excess of cadmium acetate (0.2 M). Four volumes of absolute ethanol preheated at 80°C was added to each sample. After vigorous mixing, the solutions were incubated for 20 min at 80°C, then cooled in an ice bath and, finally, centrifuged at 2000 × g for 10 min. The supernatant was kept, and the pellet was mixed with 2 vol of 80% (vol/vol) cold ethanol. After centrifugation, the two supernatants were pooled and evaporated to dryness in a rotary evaporator at 40°C. The extracts were desalted on a Sep-Pak C18 cartridge. HPLC was carried out at RT by a two-pump gradient system (Waters Assoc., Milford, MA). Vitamers were separated on a C18 column (Merck, Darmstadt, Germany) by the method of Gimsing and Beck(17). The mobile phase consisted of 0.085 M phosphoric acid titrated to pH 3.0 with triethanolamine (phase A) and acetonitrile (phase B). A 10-50% linear gradient of phase B was performed in 20 min at a flow rate at 0.5 mL/min. The effluent was collected in 0.5-min fractions. The elution position of standard vitamers (OH-Cbl, Ado-Cbl, Me-Cbl, CN-Cbl) was determined as described(15,16) using concentrated solutions of standard Cbl and absorbance detection at 365 nm. The position of the eluted vitamers was determined by γ-counting.

RESULTS

In hematoxylin-eosin-stained sections, the stomach and small intestine differentiation was similar to what it would have been if the pregnancies had not been interrupted. Oxyntic cells could be detected in every gastric fundic mucosa, and their cytoplasms were stained by the antibody to IF (Fig. 1, bottom). This protein could not be detected in the chief and mucous neck cells. In in situ hybridization-stained sections of the grafted stomachs and intestines, all the epithelial cells and the smooth muscle cells of the tunica muscularis were of human origin (Fig. 1, top). On the other hand, 95% of the endothelial cells of the intestinal and gastric microvessels and all the mononuclear cells in the lamina propria were stained by the murine DNA probes. If only 2% of the fusiform mononuclear cells that corresponded to fibroblasts in the submucosa were detected by the murine probe, 98% of the same type cells in the serosa were of host origin.

Figure 1
figure 1

Top, Section of a human embryo-fetal stomach as described previously(9). Stomach was removed from the nude mouse on d 10 and the section was hybridized with a human DNA probe. All the epithelia and muscular cells are of human origin (×40). Bottom, Immunohistochemical detection of IF from a section of a human embryo-fetal gastric mucosa removed from a nude mouse on d 10. Parietal cells with detectable IF are numerous in the fundic glands (×200).

The secretion and the isoprotein pattern of IF and HC were studied in the gastric fluid aspirated from nonstimulated xenografted stomachs. The volume of 15 gastric fluids, whose pH ranged from 0.8 to 6.3, was in the order of 0.2-1.0 mL. Their aspect was comparable to that of gastric juices from adult subjects. The protein concentration ranged from 1.9 to 17.5 mg/mL. After incubation with labeled Cbl, in the presence or absence of cobinamide, the gastric juices were run in a Sephacryl S300 gel filtration column. Only IF saturated with labeled *Cbl was detected in the presence of cobinamide. The HC-Cbl peak was clearly separated from IF-Cbl in Sephacryl S300 chromatography, with retention times of 51 and 56 min, respectively (Fig. 2, top left). The 56-min peak shifted into the void volume when it was incubated with a rabbit antiserum to human IF before its analysis in gel filtration, showing that it corresponded to IF (Fig. 2, top left). By contrast, no modification of the elution profile was obtained after addition to the sample of an antiserum to human TC. No correlation was found between the pH and the concentration of IF. This lack of correlation was not because of blood contamination, as no Hb was detected in the gastric fluid. A third [57Co]Cbl-protein peak was observed at a retention time of 63 min: its elution position was the same as that of human blood TC (Fig. 2, top left). This [57Co]Cbl TC-like protein complex did not react with antisera to human IF or TC (data not shown). Its elution profile in gel filtration and its isoprotein pattern were similar to those obtained with TC from mouse blood. The concentration ranges of IF, HC, and TC-like protein were 5.0-26.4, 1.9-27.1, and 4.1-12.6 pmol/mL, respectively, and their specific activity ranged from 0.9 to 4.3, 0.2 to 13.9, and 0.2 to 1.3 pmol/mg protein, respectively (Table 1). The fractions of [57Co]Cbl-protein peaks eluted in gel filtration were run in an isoelectrofocusing column. The IF-Cbl complex appeared as a single isoprotein with a pI of 5.6 (Fig. 2, top right). The isoprotein maintained the same pI after incubation of the IF-Cbl complex with 0.5 U of neuraminidase (Fig. 2, bottom left). By contrast, the fraction corresponding to HC had two isoproteins whose pI were 4.9 and 5.3 (Fig. 2, bottom right).

Figure 2
figure 2

Top left, Sephacryl S300 gel filtration chromatography of the Cbl-binding proteins of gastric juice from human embryo-fetal stomach xenograft. Three Cbl-binding proteins were identified when gastric juice was incubated with [57Co]Cbl:HC, IF, and a TC-like Cbl-binding protein; these proteins had a respective retention time of 51, 56, and 63 min (–•–). The [57Co]Cbl-IF peak shifted into the void volume position [same retention time as dextran blue (DB) 2000] when the sample was preincubated with an anti-IF antiserum, corresponding to the formation of an immune complex (―|―-). The [57Co]Cbl-HC and [57Co]Cbl-TC-like peaks disappeared when the sample was preincubated with a 200-fold excess (in relation to [57Co]Cbl) of cobinamide (——). Top right, Isoelectrofocusing of the [57Co]Cbl-IF peak collected from Sephacryl S300 gel filtration of gastric juice preincubated with cobinamide. A single isoprotein was observed, with a pI at 5.6. Its pI was not modified when the[57Co]Cbl-IF fraction was subjected to a treatment by neuraminidase (bottom left). The isoelectrofocusing of the [57Co]Cbl-HC fraction showed two isoproteins with isoelectric points at 4.9 and 5.3, respectively (bottom right) (cpm,——; pH, —δ—).

Table 1 Concentration of unsaturated Cbl-binding proteins in gastric juice (GJ) and intestinal juice (IJ) from fetal xenografts, determined by gel filtration chromatography*

The IF-mediated intestinal uptake of Cbl was assessed either by detecting the IF-*Cbl receptor activity in mucosal homogenate or by performing Schilling tests with [57Co]-labeled Cbl instilled free or IF-bound in the fetal intestine graft. An aliquot from the IF-[57Co]Cbl collected by gel filtration was incubated with 100 µL of mucosal homogenate from one fetal intestine and was run in a Sephacryl S300 gel filtration column; in the presence of CaCl2, an IF-Cbl receptor activity of 48.4 fmol/mL (9.1 fmol/mg protein) was detected. This receptor peak was not observed in the presence of EDTA (data not shown). Schilling tests were performed by measuring the excretion of labeled Cbl in mouse urines collected by 24 h, after instillation of the tracer into the xenograft lumen (Fig. 3). Using IF-[57Co]Cbl as a tracer, the excretion percentage of labeled Cbl ranged from 4.9 to 16.1% (n = 6). This percentage decreased to 0.3-0.4% after pretreatment of the animals by chloroquine (n = 2). The excretion percentages of free [57Co]Cbl were 0.3 and 1.6% in two animals. The experiment was repeated by instilling the tracer into the lumen of the intestinal xenograft in the presence of 0.1 M EDTA, which permeabilizes the mucosa. Under these conditions, the excretion percentage increased from 0.3 to 6.9% (data not shown). In two cases of Schilling tests performed with IF-[57Co]Cbl, the tracer excreted into the urines could be analyzed by HPLC. The vitamer distribution was studied in the two urine samples collected between 0 and 6 h and 6 and 24 h. The results are summarized in Table 2. Cyano-[57Co]Cbl was the main vitamin recovered in the four urine samples, showing a low coenzyme conversion of the assimilated labeled vitamin.

Figure 3
figure 3

Schilling tests with either [57Co]Cbl-IF or [57Co]Cbl in nude mice transplanted with human embryo-fetal intestinal xenograft. The tracer (450 000 cpm per test) was instilled into the xenograft after an intramuscular injection of nonlabeled OH-Cbl in the recipient animal. Urine was collected in pockets for 24 h (see Fig. 1). The excretion percentage range of the [57Co]Cbl was at 4.9-16.1% when the tracer was instilled with IF ([57Co]Cbl-IF). It decreased to 0.3-0.4% when chloroquine (0.1 mL of a solution at 16 mg/mL) was instilled into the xenograft 1 h before the administration of [57Co]Cbl-IF. The excretion percentage was at 0.3- 1.6% when [57Co]Cbl was instilled in the absence of IF.

Table 2 Relative distribution of vitamers of [57Co] Cbl excreted into the urine in two nude mice transplanted with fetal human intestine

DISCUSSION

The development and differentiation of normal human organs is usually studied in organ culture or analogous animal models. In 1991, Winter et al.(18) showed that human fetal intestine survives transplantation into athymic mice, vascularizes, matures, and is capable of developing a mucosal immune system from the host. In the present work, we used human fetal stomachs and intestines transplanted into nude mice to study the functional capacity of the fetal stomach and intestine to mediate the uptake of Cbl, considering that this vitamin is present in amniotic fluid swallowed by the fetus(9,19). These studies were performed on transplants 6 to 8 wk after grafting. As the induced abortions were carried out between the eighth and 10th week of amenorrhea, the digestive grafts studied in this work can be considered 14- to 18-wk-old fetal stomachs and intestines. Differentiation of the grafted stomachs and intestines was similar to that which would have been reached in utero at the time of graft removal, and in situ hybridization showed that all the graft epithelial cells were of human origin, as reported elsewhere(18). IF was immunohistochemically localized to oxyntic cells that could be easily detected in every gastric graft, as previously described in mature gastric mucosa(2023).

Despite the absence of cholinergic innervation, 6 to 8 wk after transplantation, graft parietal cells not only synthesized but also secreted IF. IF was 2 times less concentrated in fluids from gastric transplants than in gastric juices of human adults, and its concentration was 20 times higher than in the human amniotic fluid after the 13th week of gestation. The isoprotein patterns of IF secreted by gastric transplant resembled IF from amniotic fluid, but not IF from adult gastric secretion. When it is purified from adult gastric secretions, IF is known to be heterogeneous with acidic isoproteins. By contrast, transplant IF was characterized by a single pI, showing a restriction of the protein microheterogeneity close to that previously observed for amniotic fluid IF collected after the 13th week of gestation. This indicates that IF of amniotic fluid may originate from fetal stomach(9). The low fetal IF heterogeneity was due to an absence of sialylation, as its incubation with neuraminidase did not modify its pI(4,2128) (Fig. 2, top right and bottom left). This low sialic acid content did not affect its activity, as the protein bound to both Cbl and its intestinal receptor. In two previous works(29,30), a decrease of the binding capacity of human IF was observed after removing its carbohydrate core by a mixture of glycosidases. In contrast, glycosylation was not required for ligand or receptor binding by a recombinant rat IF expressed in Cos-1 cells(31). Our data did not permit argument about the role of the carbohydrate core of IF on its binding function, as we only studied the sialylation of the molecule.

A third Cbl-binding protein that had the molecular size of TC II and no immunoreactivity to antihuman TC rabbit serum was detected in the graft gastric juices. In fact, as both the isoelectrofocusing patterns and the molecular sizes of the unknown protein and the mouse TC were similar, the protein may correspond to mouse TC II diffused into the graft lumen from the xenograft microcirculatory bed rather than to TC II from mouse blood, as no Hb was detected in the gastric fluids. This mouse TC II could be synthesized by the transplant endothelial cells. Indeed, endothelial cells in culture, such as the HUVE cells, are known to synthesize large amounts of TC II(32), and we showed by in situ hybridization that, at time of graft removal, part of the graft endothelial cells were of murine origin. Such a hypothesis could not be tested because, as far as we know, species-specific antibodies to mouse TC II are not available.

The IF-Cbl receptor activity in the mucosal extract was at the same level as that previously reported in human fetal intestine(2). The IF dependent uptake of [57Co]Cbl was subsequently studied by performing Schilling tests in recipient mice. The percentage of the [57Co]Cbl excreted in urines after instillation of the tracer in the grafted intestine was highly variable, as previously reported for fetal intestine. This may correspond to variations in the IF-Cbl receptor expression(2). The uptake of free [57Co]Cbl was significantly lower than that of the IF-[57Co]Cbl complex, suggesting that the uptake of Cbl was mainly mediated by the receptor binding of this complex.

The excretion rate was close to zero when chloroquine was instilled into the graft 1 h before the tracer (Fig. 3). It confirms previous data showing that chloroquine inhibits the intracellular transport of labeled Cbl in pregnant mouse intestine(33), rat intestine, and HT 29 cells(34,35), a cell line issued from human colon adenocarcinoma that resembles fetal enterocytes(36). The inhibiting effect of chloroquine may be due to an inhibition of the degradation of IF inside the cells and to a subsequent decreased transfer of Cbl to TC II(34).

The analysis of Cbl vitamers by HPLC was performed to investigate a possible conversion of the tracer into coenzymes during its passage through the xenograft mucosa. In this model, the animals were injected with nonlabeled Cbl before the instillation of the tracer into the human xenograft, and no uptake of the [57Co]Cbl by mouse tissues could theoretically have occurred. In fact, we observed that the main vitamer recovered in urines was CN-Cbl (Table 2). Part of dietary Cbl is known to be converted into coenzymes (Me- and Ado-Cbl) in the enterocyte(37). It may be assumed, therefore, that CN-Cbl undergoes a dual fate in the digestive fetal mucosa, either sequestered as coenzymes within the enterocytes or directed to the blood stream in the cyano form. Such a dual fate was previously shown by our group, by studying the transcytosis of labeled Cbl and coenzyme distribution in CaCo2 cells cultivated on filters(38).

Fetal swallowing is one of the mechanisms of amniotic fluid circulation(19) and contributes greatly to amniotic fluid resorption, reaching 500 to 1000 mL/d near term(39). It may contribute to fetal somatic and gastrointestinal development. Recently, a number of defects in enterocyte morphology, such as glycogen assimilation and altered lysosomal morphology, have been observed in the gastrointestinal tract of fetal sheep in which the fetal ingestion of amniotic fluid was prevented(40). HC, TC, and IF have been found in amniotic fluid(9,41). Our data showed that IF from amniotic fluid may originate from the fetal stomach and that IF secreted by the fetal stomach can mediate Cbl absorption in the fetal digestive tract. Because IF receptor is expressed in the proximal as well as in the distal fetal intestine before the 15th week of gestation, one may assume that the uptake of Cbl is effective along the whole intestine and participates in the delivery of Cbl to the fetus. Further experiments are needed to assess whether such an uptake plays a role in the differentiation and development of the fetal intestine.

In conclusion, we showed that the fetal stomach secretes a biologically active IF that resembles amniotic fluid IF and that potentiates the Cbl uptake by fetal intestine mucosa. It is likely, therefore, that Cbl from amniotic fluid can be absorbed by the fetal intestine in vivo and that it is one of the possible sources of Cbl for the fetus.