Main

VEGF, also known as vascular permeability factor (VPF) or vasculotropin, is a homodimeric 34-42-kD heparin-binding glycoprotein(1) with potent angiogenic, mitogenic, and vascular permeability-enhancing activities. The gene for human VEGF is organized into eight exons(2). As a result of alternative RNA splicing from a single gene, four transcripts encode mature monomeric VEGF containing 121, 165, 189, and 206 amino acid residues (VEGF121, VEGF165, VEGF189, VEGF206). VEGF121 is a soluble form whereas VEGF165 is partially membrane bound and partially secreted. VEGF189 and VEGF206 both have a high affinity for heparin and they also bind to heparin-containing proteoglycans in the extracellular matrix(3). However, VEGF165 is the predominant isoform in most systems that have been studied(4). A large number of cell types express active VEGF. It is expressed by rodent and human tumor cells as well as by normal tissues including the epithelial cells of lung alveoli, adrenal cortex, kidney, and breast(5). It is also present in several biologic fluids such as blood, joint fluid, aqueous, and in the vitreous humor(4). Its concentration increases with inflammation and vascular proliferation. Examples of clinical conditions in which VEGF is increased include tumor ascites, rheumatoid arthritis, diabetes, and retinopathy of prematurity. No reports exist of its concentration in breast milk.

Vascular endothelium expresses two high affinity tyrosine kinase receptors for VEGF. These are the 160-kD fms-like tyrosine kinase (flt-1) and the 190-kD kinase insert domain-containing receptor (kdr) and its murine homologue fetal liver kinase (flk-1). On Scatchard analysis, the kd of the receptor flt-1 for endothelial cells has been estimated at 1-20 pM and that for kdr/flk-1 between 50 and 770 pM(4). Both flt-1 and receptors kdr/flk-1 (or their mRNA) have been found in nonendothelial cells(5). In addition to these receptors with high affinity to VEGF, receptors with much lower affinity may also exist. Cross-linking experiments have identified an unrelated 120/130- kD receptor complex with a relatively low kd of approximately 20-30 nM that binds the VEGF165 isoform but not the VEGF121. This coreceptor is identical to Neuropilin-1(6). However, the presence of VEGF receptors on human enterocytes has not so far been reported.

The physicochemical environment of the intestine is derived from dietary intake, bacterial ecology, and glandular secretions. In the nursing neonate, breast milk is an important regulator of that environment(7). The lower incidence of necrotizing enterocolitis among breast-fed infants may be due to factors present in breast milk(8). We hypothesized that breast milk, a biologic fluid rich in cytokines(9), growth factors, and antibacterial and immunomodulating agents, may contain VEGF. The consistency of breast milk varies with gestational age and with duration of lactation. We, therefore, examined the effect of these two variables on VEGF concentration. Further, to assess whether VEGF may have a biologic effect on the newborn mucosa, we examined for the presence of VEGF receptors on the apical surface of enterocytes. The intestinal epithelial line (Caco-2) was used as it is a recognized model of the small intestinal epithelium(10). Cultured Caco-2 cells spontaneously differentiate into cells with features similar to enterocytes in vivo. Changes in differentiation are detectable by measuring brush border enzyme activity(11) and by examining polarity in morphologic studies.

METHODS

Cells and culture. Caco-2 cells (American Type Culture Collection, Rockville, MD) were maintained in 10% FCS (Hyclone, Logan, UT) containing DMEM [supplemented with 2 mM glutamine, 0.1 mM nonessential amino acids, 100 000 U/L penicillin, 100 mg/L streptomycin, and 10 mM HEPES (GIBCO, Grand Island, NY)]. Cells were used between the 20th and 25th passages. HUV-EC-C were used as positive controls for VEGF-induced proliferation and for the detection of mRNA of VEGF receptors. They were maintained in 90% F12K medium (GIBCO), 10% FCS, 100 µg/mL heparin (Sigma Chemical Co., St. Louis, MO), and 30 µg/mL ECGS (Becton Dickinson Labware, Bedford, MA). Cells were used between the 4th and 7th passage. Fibroblasts, which served as negative controls in the detection of flt-1 mRNA, were grown from human fetal skin(12). They were maintained in 90% F12 medium (GIBCO), 10% FCS, 100 000 U/L penicillin, and 100 mg/L streptomycin. All cells were grown in a humidified atmosphere at 37°C with 5% CO2.

Human breast milk. Fresh human breast milk was obtained from healthy mothers of premature and full-term babies. All mothers had planned to breast-feed and volunteered to participate in the study. The donors were recruited from the obstetric wards of Massachusetts General Hospital, Boston, MA and the nursery of Aghia Sophia Children's Hospital, Athens, Greece. The milk was collected via a low-pressure manual pump (Medela Inc., P.O. Box 660, McHenry, IL). Breast milk samples were centrifuged immediately at 680 × g for 10 min at room temperature. The cells were then separated and the lipid layer was discarded. The aqueous phase was collected and stored at -80°C before analysis.

Quantification of VEGF. The concentration of monomeric VEGF165 was measured in the aqueous phase of human milk by ELISA according to the manufacturer's instructions (Quantikine, R&D Systems Inc., Minneapolis, MN). Anti-VEGF antibody reactivity was quantified spectrophotometrically at 450 nm. Serum concentrations were also measured to verify that the VEGF results from this assay corresponded to those reported in other centers(13).

VEGF receptor binding assay. Binding assays were performed as described by Oguchi et al.(11) Briefly, Caco-2 cells were grown to confluence in 48-well plates. At d 6 (less-differentiated cells) and 21 (well-differentiated cells), they were incubated at 24°C for 3.5 h in binding buffer [DMEM plus 0.1% BSA (low endotoxin BSA, Sigma Chemical Co.)] at a density of 8 × 105 cells per well with increasing concentrations (25, 50, 75, and 100 pM) of 125I-VEGF165 (Biomedical Technologic Inc., Stoughton, MA; specific activity of 2790 Ci/Mmol). Nonspecific binding was determined by the addition of 100-fold excess of unlabeled recombinant human VEGF165 (Pepro Tech Inc., Rochy Hill, NJ). For competition assays, the cells were incubated with a constant concentration of 125I-VEGF165 (50 pM) and increasing amounts of unlabeled VEGF165. After adding the labeled and unlabeled VEGF165, the final volume was 200 µL per well. Cells were then washed three times with binding buffer and solubilized with 1 N NaOH. Radioactivity present in the lysates was quantified in triplicate (three separate wells) using a gamma counter (1275 Mini Gamma LKB, Wallac, Turku, Finland). Preliminary experiments showed that the binding reached a plateau after a 3-h incubation at 24°C. The specific binding was obtained by subtracting the nonspecific binding from the total binding.

Cell proliferation assay. Caco-2 cells were subcultured in 6-well plates (Falcon, Lincoln Park, NJ) at an initial density of 5 × 104 cells/cm2. Cells were maintained in DMEM with FCS. FCS was removed on d 3. From d 4, cells were cultured in media with and without human recombinant VEGF165 at 1, 25, and 50 nM. Fresh experimental media (0.5 mL per well) were replaced every 2 d. Cells were removed using 0.255 trypsin/L mM EDTA (GIBCO) and cell numbers were counted in triplicate using a hematocytometer on d 3, 5, 7, 9, 11, and 13. Control experiments were performed with HUV-EC-C. They were subcultured in 6-well plates at an initial density of 5 × 104 cells/cm2. On d 2, the medium changed from maintenance medium (as described above) to 98% FK12, 2% FCS without ECGS and heparin and on d 3 to FK12. VEGF165 at the above concentrations was added in 50% of the wells on d 4. The proliferative effect was assessed by hematocytometer on d 5.

RNA isolation. Cells were collected from 75-cm2 flasks via trypsinization, the suspension was centrifuged for 3 min at 800 rpm, and the pellet was dissolved into at least 20-vol GIT buffer [4 mol/L GIT, 50 mmol/L Tris pH 7.6, 2% Sarkosyl (Fluka, Switzerland), and 100 mmol/L 2-mercaptoethanol]. The RNA were deproteinized by selective precipitation from GIT buffer and by extraction with phenol, chloroform, and isoamyl alcohol(14). RNA was quantified by absorbance at 260 nm and stored at -20°C. The ratio of A260/A280 was >1.8.

RT-PCR. Total RNA (1 µg) was used to synthesize cDNA in a total volume of 20 µL with 2.5 U/µL reverse transcriptase (RT). cDNA were amplified by PCR in 100 µL using the Gene AMP RNA PCR kit (Perkin-Elmer, Roche Molecular Systems Inc., Branchburg, NJ) according to the manufacturer's recommendations. The following PCR primers for the VEGF receptor flt-1 were used: upstream primer: CTAGGATCCGT-GACTTATTTTTTCTCAACAAGG; downstream primer: CTCGAATTCAGATCTTCCATAGTGATGGGCT(15). The PCR primers for kdr/flk-1 receptor were a gift from Dr. Dvorak's laboratory. Amplifications were performed in a Hybaid thermal reactor (National Labnet, NJ) under the following conditions: 35 cycles; annealing temperature 62°C; extension for 2 min at 72°C; melting temperature 95°C for 30 s. PCR products (50 µL) were separated by electrophoresis through a 1.0% agarose gel containing ethidium bromide and visualized by UV illumination. HUV-EC-C served as positive controls and human fetal skin fibroblasts as negative controls for the detection of mRNA of flt-1 VEGF receptor. Control glyceral-dehyde phosphate dehydrogenase (GAPDH) mRNA was examined in Caco-2 cells, endothelial cells, and fibroblasts.

Statistics. Comparisons were made using the t test. A. probability of <0.05 was considered significant.

RESULTS

VEGF present at high concentration in the aqueous phase of human milk. Monomeric VEGF containing 165 residues was examined by ELISA in the breast milk of 43 mothers up to 5 mo postpartum. Twenty mothers delivered prematurely (gestational age <37 wk) and 23 were full-term pregnancies. The concentration of VEGF165 ranged between 9.5 and 149.5 ng/mL. VEGF in the aqueous phase of human milk is, therefore, 2 orders of magnitude greater than in serum(13).

To examine the effect of gestational age on VEGF concentration of human milk, we compared VEGF165 concentration between mothers of full-term (n = 13) and mothers of preterm (n = 20) infants in the first week of lactation. The mothers of preterm infants were assigned to two groups according to their gestational age. Seven infants were born at gestational age of 26-31 wk and 13 were 32-36 wk. In addition, 13 mothers delivered at term (Fig. 1). The mean VEGF concentration of the group of mothers who had full-term pregnancy was greater than the means of groups whose gestational ages were 26-31 or 32-36 wk (p < 0.05 and p < 0.01, respectively).

Figure 1
figure 1

Breast milk VEGF concentrations (ng/mL) increase with gestational age. Breast milk was collected from healthy mothers of premature 26-31 wk (n = 7), 32-36 wk (n = 13) gestational age, and full-term infants (n = 13). All samples were obtained within the first week after delivery. VEGF165 concentration in breast milk was measured by ELISA. The mean VEGF165 concentration in the breast milk of mothers who delivered at term was greater than in the groups whose gestational ages were 26-31 (p < 0.05) or 32-36 wk (p < 0.01). There was no statistically significant difference between the mean VEGF165 breast milk concentration in the groups of mothers who delivered prematurely. The bars represent standard deviations of the mean VEGF165 breast milk concentration of mothers of the three different groups.

Full size image

To examine the effect of duration of lactation on VEGF concentration, we compared the VEGF165 concentration in the breast milk of mothers of full-term infants during the first week of lactation (n = 13) with the second week (n = 7) (Fig. 2). VEGF concentrations were greater in the first week of life (p < 0.01). Finally, later in lactation (2.5, 4, and 5 mo), we measured the breast milk VEGF165 concentration in mothers who gave birth to term babies. The range was 22.7 to 35 ng/mL.

Figure 2
figure 2

VEGF in breast milk decreases with duration of lactation. Samples were obtained from mothers who delivered at term (gestational age 37-41 wk) in the first (n = 13) and second postnatal week (n = 7). VEGF165 concentration in breast milk was measured by ELISA. VEGF165 concentration was significantly higher in the first week of lactation than in the second (p < 0.01). There was also a statistically significant fall in the concentration after the first week (data not shown). The bars represent standard deviations of the mean breast milk concentration in the two groups.

Full size image

VEGF binds specifically to human enterocytes. In four separate experiments in Caco-2 cells, we observed a specific binding of VEGF165 ranging from 78 to 92% of total binding (Fig. 3). The mean was 86.2% with SEM of 1.67%. Scatchard analysis of the specific binding data showed the presence of binding sites for VEGF165 on the surface of Caco-2 cells (Fig. 4). The kd of VEGF receptor in Caco-2 cells was estimated to be 2.85 to 4 nM, which is higher than that observed in endothelial cells. The specificity of the receptor for VEGF165 was confirmed using radiolabeled 125I-VEGF165 in four competitive binding experiments with unlabeled VEGF165 (Fig. 5). To examine the change in the number and affinity of VEGF receptors with cell differentiation, experiments were performed on both less-differentiated (d 6) and well-differentiated (d 21) cell monolayers. Caco-2 cell differentiation had no effect on VEGF receptor affinity or density (data not shown). VEGF receptor mRNA was examined by RT-PCR. Using the primers described in "Methods," we detected a 225-bp product corresponding to flt-1(15) (Fig. 6). Flt-1 mRNA was not sufficiently abundant to be detected by Northern blot hybridization. In addition, the mRNA of kdr/flk-1 was not present in Caco-2 cells in three experiments.

Figure 3
figure 3

125I-VEGF165 binds specific cell surface receptors on Caco-2 cells. Receptor-ligand binding assay was performed on 8 × 105 cells in 48-well plates for 3.5 h at room temperature in the presence of increasing concentrations of 125I-VEGF165 (25, 50, 75, 100 pM). The nonspecific binding was determined by adding 100-fold excess of unlabeled VEGF165. The specific binding of VEGF165 ranged from 78 to 92% of the total binding. Each point is the mean of three determinations and representative of four separate experiments.

Full size image
Figure 4
figure 4

Scatchard analysis of specific binding data of 125I-VEGF165 to Caco-2 cells. Scatchard plot shows the presence of a single class of high affinity binding sites for VEGF. The kd ranged from 2.85 to 4 nM. Each point is the mean of triplicate measurements and representative of four separate experiments.

Full size image
Figure 5
figure 5

Competition of 125I-VEGF165 binding to Caco-2 cells by unlabeled VEGF165. Binding was performed on 8 × 105 cells in 48-well plates for 3.5 h at room temperature in the presence of a constant concentration of 125I-VEGF165 (50 pM) and increasing concentrations of VEGF165. The curve shows inhibition of 125I-VEGF165 binding to Caco-2 cells by unlabeled VEGF165. Each point is the mean of triplicate measurements and representative of five separate experiments.

Full size image
Figure 6
figure 6

Flt-1 mRNA is present in Caco-2 cells. The transcripts for the VEGF receptor were examined by RT-PCR using flt-1 specific primers. Caco-2 cells and endothelial cells contained flt-1 mRNA. However, this transcript was not present in fibroblasts. GAPDH mRNA was detected in all three cell types. (Representative of seven experiments with similar results.)

Full size image

VEGF has no proliferative effect on Caco-2 cells. We evaluated the mitogenic effect of VEGF on intestinal epithelial cells. Caco-2 cells were cultured in DMEM supplemented with various concentrations of VEGF165 (1, 25, and 50 nM). Although VEGF165 induced proliferation on endothelial cells within 24 h, it had no effect on intestinal epithelial cell proliferation even after prolonged incubation (Table 1).

Table 1 VEGF has no effect on intestinal epithelial cell proliferation
Full size table

DISCUSSION

The present study describes the first demonstration of VEGF in breast milk. Its immunochemical concentration (9.5 to 149.5 ng/mL) was 2 orders of magnitude greater than in the serum of normal adults (range 0 to 0.227 ng/mL)(16). In addition, the breast milk concentration of VEGF varied with gestational age of pregnancy and with lactation. We speculate that the changes in VEGF concentration may be due to the recognized hormonal changes in pregnancy. VEGF increased toward the end of pregnancy when levels of estrogen and progesterone are greatest; both hormones are known to up-regulate VEGF mRNA in stromal and granular cells in rat uterus(17). Further, the fall in VEGF concentration observed 1 wk postpartum occurred as the serum level of these hormones decreased after birth. The high concentration in breast milk during the first days postpartum also coincides with the high numbers of cells in colostrum(7).

The concentration of VEGF in human milk in the present study was the same order of magnitude as the kd of its receptor calculated from our experiments on Caco-2 cells. The concentration of bioavailable VEGF reaching the human intestinal surface in vivo is difficult to assess. Enzymes in the stomach and duodenum may digest growth factors. Nevertheless, the high buffering capacity of milk may, to some extent, protect the growth factors from degradation. The kd of VEGF binding to the cell line was 4- to 165-fold greater than that observed for the high affinity receptors flt-1 and kdr/flk-1 in endothelial cells(18) and 10-fold lower than the kd of low affinity VEGF receptors(3). The lower affinity of Caco-2 cell receptor for VEGF compared with flt-1, kdr/flk-1 endothelial receptors may be due to a difference between the environment at the apical surface of the enterocyte and that of an endothelial cell. In vivo the passage of molecules from the bulk phase of the intestinal lumen to the epithelial cell apex encounters two specific regions, the unstirred layer and the acid microclimate, both of which are present in the neonatal intestine(19). The thickness of unstirred layer, an important component of which is mucus, represent a significant barrier to molecules reaching the apical surface. Mucus present at the apical surface can trap macro-molecules such as VEGF. Mucus covering the apical VEGF receptors would hinder their interaction with VEGF molecules(20). The acid microclimate is the result of the apical Na/H+ exchange pump function. Hydrogen ions are pumped out of the cells and trapped by the negatively charged mucopolysaccharide side chains. Its effects are greatest during lactation. In the suckling rat, the microclimate is more acidic than in the adult(21). VEGF, upon reduction, denatures and loses its biologic activity(4,22). Thus, the acid microclimate may reduce the activity of VEGF, explaining the lower affinity observed in our experiments than in those of endothelial cells that do not maintain an acidic microclimate.

The mRNA for the flt-1 VEGF receptor for endothelial cells was readily detected by RT-PCR(14). Northern blot hybridization experiments do not show an abundant signal in endothelial cells(15). This was probably due to the fact that Northern analysis is not sufficiently sensitive to detect transcripts in low copy number. In our experiments, Caco-2 cells contained flt-1 mRNA, as detected by RT-PCR. VEGF had no mitogenic activity upon Caco-2 cells. This finding is in keeping with detection of flt-1 mRNA. Previous studies(18) have shown that this receptor does not induce mitogenic signaling in flt-1-expressing porcine endothelial cells nor in National Institutes of Health 3T3 cells transfected with flt-1 cDNA. Its effects are limited to molecular fluxes into cells.

The present study reports experiments in which the model of intestinal epithelium was the Caco-2 cell line. Binding experiments have also been performed on the H-4 cell line(12), a nonmalignant epithelial cell line derived from human fetal small intestine (data not shown). We used cells between the 9th and 12th passage. The kd of VEGF receptor in these cells was calculated at 1.55-2.66 nM, which was similar to that observed in Caco-2 cells. VEGF165 had no mitogenic activity on H-4 cells, again in keeping with our observations on Caco-2 cells. Thus, the observations made on Caco-2 cells were not limited to that cell type alone. The potential biologic role of VEGF upon the intestinal mucosa is currently unknown. Although it does not affect cell proliferation, it may enhance the glucose entry in the enterocyte, an important function in a cell under stress, by up-regulating the glucose transporter genes as it does for the endothelial cell(23). It also may have an immunomodulatory effect on the newborn's intestinal mucosa by altering the permeability of the epithelium to macromolecules.

In conclusion, VEGF is present in high concentrations in human milk. It binds to specific cell surface receptors found on the apical surface of intestinal epithelial cells. These receptors are likely to be of the flt-1 type because flt-1 receptor mRNA was detected by PCR. Interaction between VEGF and its receptor in intestinal epithelial cells may, therefore, occur in the newborn infant.