点击显示 收起
1 From the Departments of Obstetrics and Gynecology and Pediatrics, Helsinki University Central Hospital, and the Department of Pathology, Haartman Institute, Helsinki, Finland.
2 Supported by the Clinical Research Fund of Helsinki University Central Hospital, the Finnish Foundation for Obstetric and Gynecologic Research, and the Finnish Diabetes Research Fund. 3 Reprints not available. Address correspondence to P Vuorela, Research Laboratory, Department of Obstetrics and Gynecology, Helsinki University Central Hospital PL 140, 00029 HYKS, Helsinki, Finland. E-mail: piia.vuorela{at}kolumbus.fi.
ABSTRACT
Background: Human milk, rich in cytokines, may contain the potent permeability- and angiogenesis-promoting agent vascular endothelial growth factor (VEGF).
Objective: We wanted to study whether free or bound VEGF is present in human milk and whether it and its receptors (VEGFR-1 and -2) are expressed in lactating breast or newborn intestinal tissue.
Design: The study had a longitudinal design with collection of human milk from healthy (n = 32) and diabetic (n = 5) women at 2, 7, and 30 d postpartum. Milk was analyzed for VEGF by enzyme-linked immunosorbent assay along with plasma samples collected 2 d postpartum. Immunohistochemistry was used to localize VEGF and its receptors in lactating breast and newborn intestine. Gel filtration with radiolabeled VEGF was performed to study whether human milk contains VEGF binding proteins.
Results: Human milk VEGF concentrations in healthy (76 ± 19 µg/L,
Key Words: Colostrum • human milk • vascular endothelial growth factor • VEGF • VEGF receptors • lactating breast • intestinal villi • newborn intestine • type 1 diabetes mellitus
INTRODUCTION
Human milk furnishes optimal nutrition for newborns. In addition, it protects against infections like otitis media, respiratory syncytial and rotavirus infection, and autoimmune diseases such as type 1 diabetes mellitus, as well as against atopic skin diseases (1–4). These health benefits are explained at least in part by the immunoglobulins (Igs), especially secretory IgA, as well as lactoferrin and lysozyme of human milk. Human milk contains various growth factors, such as insulin, insulin-like growth factor, transforming growth factor, epidermal growth factor, cortisol, polyamines, and several interleukins (5–12), which are thought to play an additional role in human milk's protective effects against infant infections.
Vascular endothelial growth factor (VEGF) is an endothelial cell mitogen that, in addition, induces vascular permeability and promotes monocyte migration (13). In mice, VEGF is crucial for vascular development and embryonic survival. The lack of even one allele results in intrauterine death (14, 15).
The action of VEGF is mediated by its 2 known receptors, VEGFR-1 [or FMS-like tyrosine kinase 1 (Flt-1)] and VEGFR-2 [or kinase insert–domain-containing receptor (KDR)]. The soluble form of VEGFR-1 has been identified in amniotic fluid and in serum of pregnant women, where it binds circulating VEGF (16, 17).
Milk from diabetic women may have higher glucose concentrations than does that of healthy women, and in the former, the mammary gland lipid metabolism may be impaired (18, 19). Whether diabetes affects the various cytokines in human milk is not known. However, advanced glycation end products, which are linked with diabetic complications, are known to enhance VEGF messenger RNA and secreted protein in human retinal pigment epithelial cells (20).
The presence of VEGF in breast milk was shown previously (21). We wanted to learn whether human milk contains free or bound VEGF, its possible location in lactating breast tissue, and the occurrence of VEGF receptors in the newborn gut. We also wanted to study whether human milk VEGF concentrations would be affected by diabetes because diabetes has been associated with changes in VEGF expression. For example, increased VEGF concentrations have been observed in the ocular fluid of patients with diabetic retinopathy as well as in the kidneys of patients with glomerular kidney disease (22, 23).
SUBJECTS AND METHODS
Subjects and sample collection
After permission from the ethics committee of the Helsinki University Central Hospital Department of Obstetrics and Gynecology, 34 healthy volunteer mothers (Table 1) were recruited for this longitudinal study from the postnatal wards on the day after delivery. On the second postpartum day, peripheral blood plasma samples were collected into glass tubes with EDTA; centrifuged at 3000 x g, for 10 min, at room temperature; and the plasma separated for storage at -20°C. Five milliliters of human milk was collected on the second postpartum day (n = 34) and at 7 (n = 30) and 30 (n = 27) d postpartum. Because the consistency of human milk changes during suckling, the latter samples included 5 mL of foremilk and 5 mL of hindmilk from the same breast during one feeding. Milk samples were collected into polypropylene tubes by hand expression between 0600 and 1200.
View this table:
TABLE 1. Clinical data of the study subjects
Mothers with type 1 diabetes mellitus were recruited in a similar way, but because of preterm labor and consequent problems in lactation, only 5 mothers could give a human milk sample on the second postpartum day, and only 3 of these 5 could give a sample at 7 and 30 d postpartum. All diabetic mothers received individual insulin therapy and were in good metabolic control as assessed by the percentage of glycated hemoglobin (Hb A1c) and repeated blood glucose measurements. The completeness of breast-feeeding as well as the duration of pregnancy varied between subjects and should be regarded as confounding factors in this study.
All samples were immediately frozen and stored at -20°C until analyzed. Before assessment, milk samples were centrifuged at 1000 x g at 4°C for 10 min. The aqueous fraction was collected by careful aspiration with a thin needle through the surface fat layer.
Enzyme-linked immunosorbent assay
Concentrations of free VEGF in human milk and plasma were determined by enzyme-linked immunosorbent assay (ELISA) according to manufacturer's instructions (R&D Systems, Abingdon, United Kingdom). The assay recognizes only free VEGF, and bound protein remains undetected (17). The aqueous fraction of the milk samples was diluted 50-fold with calibrator diluent provided by the assay. All samples and standards were analyzed in duplicate. Recovery of VEGF was tested by the ability of the ELISA to detect added human recombinant VEGF (200 ng/L; standard protein provided by the assay) in the samples.
We showed previously that the ELISA for VEGF does not cross-react with the related growth factors VEGF-B (24) or VEGF-C (17, 25), but according to the manufacturer, it shows 20% cross-reactivity with the VEGF-PlGF heterodimer (26).
Radioiodination of VEGF
Five micrograms of recombinant human VEGF165 (Genzyme Diagnostics, Cambridge, MA) was radiolabeled with Iodo Gen (Pierce, Rockford, IL) as described previously (17).
Gel filtration
A Sephacryl S-300 h (Pharmacia Biotech, Uppsala, Sweden) column (84 x 1.5 cm, Pharmacia LKB Biotechnology, Uppsala, Sweden) was equilibrated with tris-buffered saline (TBS; 0.05 mol tris-HCl/L, 0.15 mol NaCl/L, pH 7.7). The exclusion volume of the column was 40 mL; the flow rate was 15 mL/h (Microperpex peristaltic pump; Pharmacia LKB Biotechnology), and 1-mL fractions were collected (Redifrac; Pharmacia LKB Biotechnology) at 4°C. Ferritin (440 kDa; Pharmacia Fine Chemicals, Uppsala, Sweden), human IgG (168 kDa; Sigma Chemical Co, St Louis), bovine serum albumin (67 kDa; Sigma Chemical Co), and soybean trypsin inhibitor (20.1 kDa; Sigma Chemical Co) were used as molecular-weight markers. Fractions were monitored for absorbance at 280 nm (Lambda 3B UV/VIS spectrophotometer; Perkin-Elmer, Überlingen, Germany).
One hundred milliliters [125I]VEGF (80000 cpm) in TBS was loaded onto the column either alone or after 1 h of incubation at room temperature with 100 mL of a pool (n = 12) of the maternal plasma samples or a pool (n = 12) of human milk from each of the time points studied. The fractions were monitored for radioactivity (1260 Multigamma gamma counter; LKB, Wallac, Sweden).
Tissue samples from breast and intestine
Paraffin-embedded samples of lactating breast (n = 5) and newborn intestine (n = 3) were used for immunohistochemistry after permission from the local ethics committee. Breast samples were collected at biopsies of palpable tumors performed for diagnostic purposes. The tumors were verified as lactating adenoma, ie, a benign tumor caused by occlusion of mammary ducts leading to retainment of secreted milk. Histologically normal samples of intestine of newborns were resected at surgery done because of ileal obstruction (n = 2; 1 aged 1 wk, born at gestational age 30 wk and 1 aged 2 wk, born at gestational age 25 wk) or on closure of a stoma done because of earlier necrotizing enterocolitis (n = 1; aged 2 mo, born at gestational age 28 wk). These infants had not received human milk for 1 wk, but instead received cow milk–derived infant formula, which was also assessed for VEGF by ELISA.
Immunohistochemistry
Five-micrometer tissue sections were dewaxed, rehydrated, and incubated in 0.5% pepsin (Merck, Darmstadt, Germany) in phosphate-buffered saline (PBS; 0.14 mol NaCl/L, 2.7 mol KCl/L, 0.01 mol Na2HPO4/L, 1.76 mol KH2PO4/L, pH 7.4) at 37°C for 15 min and then in 0.6% H2O2 (Perhydrol; Merck) in methanol for 15 min at room temperature before staining.
The sections were blocked with normal swine serum and reacted with polyclonal rabbit antibodies raised against a peptide corresponding to amino acids 1–191 of VEGF of human origin (5 mg/L) and a peptide corresponding to an amino acid sequence mapping at the carboxy terminus of the precursor form of VEGFR-1 of human origin (2.5 mg/L), or they were blocked with normal rabbit serum and reacted with monoclonal mouse antibody raised against a recombinant protein corresponding to amino acids 1158–1345 mapping at the carboxy terminus of the VEGFR-2 precursor of mouse origin (2.5 mg/L). Primary antibodies, all of which recognized human antigens and were purchased from Santa Cruz Biotechnology (Santa Ana, CA), were incubated at 4°C overnight. Sections were then incubated with biotinylated swine anti-rabbit antibody (1.8 mg/L), or rabbit anti-mouse antibody (5.5 mg/L; Dako A/S, Glostrup, Denmark) at room temperature for 1 h. Between all steps, except between blocking serum and primary antibody, sections were washed 3 times for 5 min each in PBS.
Protein localization was visualized by use of an avidin-biotin-complex (Vectastain ABC kit; Vector Laboratories, Burlingame, CA) and aminoethylcarbazole (Sigma Chemical Co). Sections were counterstained with Mayer's hemaluminum (Merck), and Aquamount Improved (BDH Laboratory Supplies, Poole, United Kingdom) was used as the mounting medium.
Polyclonal rabbit antibodies against von Willebrand factor (14 mg/L; Dako A/S) were used to identify blood vessels. PBS and nonimmune serum served as negative controls and full-term placenta, a tissue known to express all the proteins studied (16, 27), served as a positive control.
Statistical analysis
Within-group comparisons of VEGF concentrations in human milk were performed by using the nonparametric Wilcoxon test, and between-group comparisons were made by using the Mann-Whitney U test. The P values were Bonferroni adjusted. Associations between different variables were examined by performing Spearman's rank correlations. The STATVIEW program (version 4.1; Abacus Concepts, Berkeley, CA) was used for the analyses.
RESULTS
VEGF in the aqueous fraction of human milk and maternal plasma
Because VEGF concentrations in the nondiabetic mothers did not differ between 10 paired foremilk and hindmilk samples collected at 1 wk postpartum, only the foremilk samples were analyzed.
We observed high VEGF concentrations in colostrum (76 ± 19 µg/L,
FIGURE 1. . Vascular endothelial growth factor (VEGF) concentration in breast milk of healthy and diabetic mothers 2 (colostrum), 7, and 30 d postpartum. Horizontal lines indicate mean values for healthy subjects. *Significantly different from other time points, P < 0.001.
The VEGF concentrations of milk of diabetic mothers did not differ significantly from those of healthy women; they were 75 ± 25 µg/L on the second postpartum day, 27 ± 8 µg/L at 7 d postpartum, and 17 ± 7 µg/L at 30 d postpartum (Figure 1).
To determine whether human milk contains any factors binding VEGF, the recovery of added VEGF in the human milk samples was analyzed. No interference of human milk with the ELISA occurred because the recovery of added VEGF, 200 ng/L, was 90–110%. Moreover, the elution profiles of radioiodinated VEGF alone or after incubation with human milk were identical. Similar results, ie, no VEGF binding, were obtained for the maternal plasma samples (Figure 2).
FIGURE 2. . Gel filtration elution profiles of [125I]vascular endothelial growth factor (VEGF) alone and after incubation with a pool (n = 12) of breast milk or maternal plasma taken 2 d postpartum. Molecular weight markers indicated by arrows: ferritin, 440 kDa; human immunoglobulin G, 168 kDa; bovine serum albumin, 67 kDa; and soybean trypsin inhibitor, 20.1 kDa.
Maternal plasma concentrations of free VEGF fell below the detection limit of 16 ng/L in all but 3 samples, in which the VEGF concentrations ranged from 23 to 39 ng/L. Thus, we could not measure any correlation between maternal plasma and human milk VEGF concentrations or other clinical measures. The cow milk–derived infant formula showed no VEGF immunoreactivity in ELISA.
Immunolocalization of VEGF and VEGFR-1 and VEGFR-2
By immunohistochemical analysis of VEGFR-1 and VEGFR-2, positive staining was evident in the luminal epithelial cells of the infant intestinal villi; VEGFR-2 was also evident in the epithelial cells of the crypts. Epithelial cells remained negative for VEGF and the villous stroma negative for all 3 proteins. In the vascular endothelium of mesenteric blood vessels, VEGF, VEGFR-1, and VEGFR-2 showed positive staining (Figure 3).
FIGURE 3. . Immunolocalization of vascular endothelial growth factor (VEGF), VEGF receptor 1 (VEGFR-1), and VEGFR-2 in newborn intestine. VEGF (A, B, C), VEGFR-1 (D, E, F), and VEGFR-2 (G, H, I) were reacted with specific antibodies; nonspecific immunoglobulin G served as a negative control (J, K, L). Epithelial cells (Ep) on apical ends of the intestinal villi were negative for VEGF (A), whereas VEGFR-1 was expressed especially on the apical surface (D). VEGFR-2 was found in the whole-cell layer (G). Epithelial cells within the crypts were negative for VEGF (B) and VEGFR-1 (E), but positive for VEGFR-2 (H). Vascular endothelium (En) in the mesenterium was positive for VEGF (C), VEGFR-1 (F), and VEGFR-2 (I). Lu, lumen. Scale bar = 100 µm.
In lactating breast, VEGF, VEGFR-1, and VEGFR-2 immunostaining was observed in the glandular epithelium of the collecting ducts. VEGFR-2 staining was also observed in the glandular epithelium of the acinal structures, whereas these were negative for VEGF and VEGFR-1. In the breast vascular endothelium, immunoreactivity for VEGF and VEGFR-2 was apparent, whereas staining for VEGFR-1 remained negative (Figure 4).
FIGURE 4. . Immunolocalization of vascular endothelial growth factor (VEGF), VEGF receptor 1 (VEGFR-1), and VEGFR-2 in lactating breast. VEGF (A, B, C), VEGFR-1 (D, E, F), and VEGFR-2 (G, H, I) were reacted with specific antibodies; nonspecific immunoglobulin G served as a negative control (J, K, L). Acinar epithelial cells (Ep) were negative for both VEGF (A) and VEGFR-1 (D) but positive for VEGFR-2 (G). The epithelium of the collecting ducts was positive for all 3: VEGF (B), VEGFR-1 (E), and VEGFR-2 (H). The vascular endothelium (En) was positive for VEGF (C) and VEGFR-2 (I) but negative for VEGFR-1 (F). Lu, lumen. Scale bar = 100 µm.
DISCUSSION
Human milk is known to be rich in cytokines, and we also detected VEGF at high concentrations in both colostrum and mature human milk. VEGF was found in µg/L concentrations, concentrations similar, for example, to those of macrophage-colony-stimulating factor and transforming growth factor ß1 present in human milk. In comparison, interleukin 10 and tumor necrosis factor occur at much lower concentrations (12). Varying data are available on the concentrations of IL-8, IL-6, and IL-1ß in human milk (12, 28–30).
Human milk VEGF concentrations in the aqueous fraction decreased significantly after delivery, which is in accordance with the findings of Siafakas et al (21). This may be explained simply by physiologic dilution of human milk during lactation. Also, the amount of VEGF consumed by the newborn is hard to measure in practice.
Maturation of the intestine is a process already begun during intrauterine life, and major adaptation takes place at birth. Prenatally, the fetus swallows fairly large volumes of amniotic fluid, which has been considered to affect the intestinal maturation process (31). Note that in amniotic fluid VEGF is readily bound (16, 17), whereas human milk contains an abundance of free VEGF. Our results suggest that no VEGF binding proteins are present in human milk. However, VEGF receptors may be present in human milk but cannot be detected by the binding of radiolabeled VEGF because of receptor occupancy.
The precise function of VEGF in newborns needs to be elucidated, but its local action in the gut rather than a role for VEGF in the systemic circulation is suggested by a finding of no absorption of radioiodinated VEGF by the intestine of newborn New Zealand White rabbits (S Andersson, unpublished observations, 1999).
The role of human milk–derived VEGF is not known, but babies not receiving human milk have more infectious diseases such as viral diarrheas (1). Because we showed VEGFR-1 and VEGFR-2 to be present in the intestinal luminal epithelium, and because VEGF has been shown to promote monocyte migration (13), it is tempting to speculate whether VEGF may be involved in the host defense mechanisms of the newborn intestine, perhaps in concert with the various other cytokines in human milk.
Diabetes was not associated with alterations in human milk VEGF concentrations compared with those of healthy women. Our initial observation was, however, limited by the number of samples because lactogenesis in diabetic women is often delayed (19), especially in women having cesarean deliveries. Because VEGF is a glycosylated protein, we hypothesized that hyperglycemia might affect its presence in breast milk. All of our diabetic subjects were, however, under good metabolic control. Thus, we cannot say whether concentrations of this glycosylated protein in human milk would have been affected by poorly controlled diabetes with hyperglycemia. Also, factors such as gestational age at birth and completeness of breast-feeding may be confounding factors in this study.
In the small intestine, the staining of VEGFR-1 showed an apical pattern, whereas VEGFR-2 was also observed in the crypts. This might imply that in the gut these 2 receptors mediate different kinds of responses to VEGF. It is also possible that the dividing cells in the crypts start expressing VEGFR-1 only after they have differentiated and migrated to the apical ends of the villi.
Cow milk–derived infant formula was assessed for VEGF to see whether infants whose intestinal samples were studied had received peroral VEGF. However, both the infant formula and the intestinal villi of the tissue sections remained negative for VEGF. Note that our samples, although representing histologically normal sites of the intestine, were from newborns with intestinal diseases, making it hard to make general conclusions. Of course, it is unethical to collect intestinal samples of healthy newborns. Also, we cannot exclude the possibility that VEGF may still be expressed by the intestinal villi and would be detectable by other methods, such as in situ hybridization of messenger RNA.
VEGF may play a role in the lactating breast because both of its receptors were detected in the glandular epithelium. However, VEGFR-1 was localized in the epithelium of the collecting ducts, whereas VEGFR-2 showed strong staining in the acinar epithelium. This may imply different roles for these 2 receptors in the lactating breast. VEGF itself was localized in the glandular epithelium of the larger collecting ducts. We did not study the site of origin of VEGF by in situ hybridization of messenger RNA. In addition to breast tissue, it is also possible that VEGF in human milk originates by diffusion from the maternal circulation. However, this seems unlikely because most maternal plasma VEGF concentrations on the second day postpartum were undetectable.
The vascular endothelial localization of VEGF and its receptors is well documented (14, 15, 32, 33) and, expectedly, they were observed in the endothelium of blood vessels in the breast and the intestinal mesenterium. In the mesenterium, however, regulation of VEGFR-1 and VEGFR-2 may differ because the mesenteric vascular endothelium stayed negative for VEGFR-1.
In conclusion, free VEGF is abundant in human milk, especially in colostrum. Its receptors, VEGFR-1 and VEGFR-2, are found in the breast ductal epithelium as well as in the intestinal epithelium of newborns. The exact role and function of VEGF in human milk remain, however, to be elucidated.
ACKNOWLEDGMENTS
We thank Eira Halenius and Pia Lehtinen for expert technical assistance, Antti Huittinen for expert desktop publishing and figure creation, and Carol Norris for revising the manuscript.
REFERENCES