Main

HGF was cloned in 1989(1) and is so named because of its potent mitogenic action on hepatocytes. It is also a mitogen to epithelial and endothelial cells and is synthesized by mesenchymal cells(2), thus implicating a potential role in epithelial-mesenchymal interactions(3). HGF is identical to scatter factor and lung fibroblast-derived growth factor which was isolated from an embryonic lung fibroblast cell line(4). The ability of cells to scatter in response to HGF in vitro may indicate a role in vivo in cell migration and tissue morphogenesis(5, 6). HGF mRNA and its protein are present in placental mesenchymal cells and amniotic epithelium of the placenta(7). HGF is also expressed in fetal kidney, liver, and lung(8).

The human placenta and maternal decidua are rich sources of growth factors which are involved in the regulation of growth and development of the fetus, including IGF-I, IGF-II, basic fibroblast growth factor, TGF-α, TGF-β, NGF, and platelet-derived growth facor(9–14). IGF-I cord blood levels increase with increasing gestational age and are directly related to gestational age(15). It has been suggested that the effect of other polypeptide growth factors in fetal and placental development may be mediated in part through IGF-I and IGF-II(16).

The role of HGF in human fetal growth and development is not known. Prior studies of HGF in embryogenesis and placental development have focused generally on the expression of HGF mRNA in animal tissues(8, 16, 18), or alternatively, the effect of HGF on either fetal cell(19) or stem cell lines(20–22). Recently a “knockout” mouse with no HGF expression has been developed. HGF would appear to be essential for normal placental and liver development in utero(23, 24).

We hypothesized that HGF levels in cord blood would increase with increasing gestational age and also be related to fetal growth parameters. We therefore aimed to determine the levels of HGF in human umbilical vein sera and their correlates to late fetal growth and gestation.

METHODS

Umbilical venous serum was collected prospectively at 148 deliveries at the Women's and Children's Hospital. The following data were recorded at birth: birth weight and length, placental weight, pregnancy history, medications, alcohol consumption and smoking, perinatal events, mode of delivery, and use of anesthesia. Gestational age was determined from the date of the last menstrual period and/or early ultrasound examination, at 16-18 wk. Ballard scoring was performed at birth by a pediatrician if the first two parameters appeared to be at variance with the neonate's maturational status. Maternal venous, umbilical venous, and umbilical artery serum was collected at nine deliveries (two sets of twins).

Sera from 32 healthy adults who were hospital employees aged 25.1 ± 0.4 y and 54 healthy adolescents aged 13.9 ± 1.1 y and attending a suburban high school in Adelaide were also obtained prospectively. The study was approved by the Women's and Children's Hospital Human Ethics Committee.

Serum samples were stored at -20°C and measured within 6 wk. Serum HGF was measured by ELISA using rhHGF and anti-rhHGF mouse MAb (Institute of Immunology Co. Ltd., Tokyo, Japan). Microtiter plates were coated with anti-rhHGF mouse MAb. Test samples or standards diluted in FCS were added. After a further incubation, the wells were washed, and horseradish peroxidase-conjugated mouse MAb specific to HGF (but to a different epitope) was added. After a further incubation the enzyme substrate (sodium perborate, tetrahydrate) containing the color developer (o-phenylendiamine, dihydrochloride) was added. The reaction was stopped by adding 4N-sulfuric acid. Absorbance was measured at 492 nm in a microtiter plate reader (Dynatech, Guernsey, Channel Islands, UK). Standard curves using this method display linearity between 0.1 and 1.6 ng/mL. The assay has no cross-reactivity with plasmin, lipoprotein(a), IGF-I, IGF-II, epidermal growth factor, fibroblast growth factor, platelet-derived growth factor, and TGF-β. The intraassay and interassay coefficients of variation in our laboratory are 1.9 and 10.9%, respectively at 0.4 ng/mL, 4.3 and 11.3% at 0.8 ng/mL, and 2.1 and 12.5% at 1.6 ng/mL.

Serum IGF-I was measured after acid/ethanol extraction of serum by a double antibody RIA as previously described(25).

Statistical analysis. Variables that were not normally distributed (HGF) were log transformed before analysis. Comparison between groups was performed using unpaired t test. Relationships between cord HGF and dependent variables (gestational age, birth weight, birth length, and placental weight) were determined by linear, polynomial, and multiple stepwise regression analyses, using a Systat statistical program.

RESULTS

Of the total 148 pregnancies, 94 were normal; 25 of these were premature deliveries (i.e. less than 37 wk of gestation). There were 54 pregnancies complicated by medical conditions; 23 of these were premature deliveries. Gestational diabetes (n = 19), insulin-dependent diabetes (n = 19), pregnancy-induced hypertension (n = 12), acute viral hepatitis (n = 1), hepatocellular carcinoma(n = 1), acute pyelonephritis (n = 1), and chronic glomerulonephritis (n = 1) complicated these 54 pregnancies.

Cord HGF levels were more than 3-fold higher than HGF levels in adolescent(p < 0.0001) and adult (p < 0.0001) sera(Table 1). There was a significant relationship between gestational age and cord HGF in normal pregnancies (r = 0.42;p < 0.0001) (Fig. 1).

Table 1 HGF levels in cord, adolescent, and adult serum samples
Full size table
Figure 1
figure 1

Relationship between cord HGF and gestation in normal pregnancies (n = 96, y = 2.518 - 0.219x + 0.004x2, r = 0.42).

Full size image

Within the normal pregnancy group, term neonates (37-42 wk of gestation,n = 69) had significantly higher cord HGF than preterm neonates(n = 25) [1.1 (0.78-1.45); 0.78 (0.46-1.03) ng/mL, respectively,p = 0.0007]. Correlations between cord HGF and birth weight(r = 0.23; p = 0.02), birth length (r = 0.23,p = 0.01), and placental weight (r = 0.22; p = 0.03) were not independent of gestational age. Furthermore, cord HGF did not differ between small for gestational age (less than 10th percentile for birth weight) and appropriate for gestational age (greater or equal to 10th percentile for birth weight) neonates at term. There was no relationship between cord HGF and cord IGF-I levels (n = 90). Cord IGF-I levels correlated with birth weight (r = 0.32, p = 0.003), birth length (r = 0.32, p = 0.002), and placental weight(r = 0.37, p = 0.001). These correlations were not independent of the significant relationship between IGF-I and gestational age(r = 0.82, p = 0.04).

Pregnancies complicated by medical conditions as detailed above had significantly higher cord HGF in premature deliveries compared with cord HGF in premature deliveries in normal pregnancies [0.92 (0.62-1.30), 0.69(0.41-1.0) ng/mL, respectively; p = 0.003]. There was no difference in cord HGF in term deliveries between normal and complicated pregnancies. Cord HGF levels in complicated pregnancies were not related to birth weight, birth length, or placental weight. There was no relationship between gestational age and cord HGF in pregnancies complicated by medical conditions, in contrast to normal pregnancies. No differences in cord HGF were observed between term and preterm neonates or between small for gestational age and appropriate for gestational age babies in complicated pregnancies. Cord HGF was not related to a history of smoking or alcohol consumption during pregnancy, mode of delivery, or use of anesthesia during labor in normal or complicated pregnancies.

Maternal venous serum HGF levels and simultaneous measurements of umbilical venous and arterial blood showed lower HGF levels in maternal serum than in the umbilical venous and arterial samples [0.56 (0.39-0.62) versus 0.68 (0.50-0.76) (p = 0.07) and 0.74 (0.47-0.89) ng/mL (p= 0.04), respectively. Umbilical venous and arterial samples were closely correlated (r = 0.825, p = 0.002, n = 11).

DISCUSSION

This study has examined HGF levels in umbilical vein sera for the first time and revealed several interesting findings. HGF levels are significantly higher in cord serum than in serum obtained from healthy adolescents and healthy adults. This is in contrast to serum IGF-I which peaks during puberty(26). There is an independent relationship between cord HGF and gestational age, but not with birth growth parameters, in normal pregnancies. However, this relationship is not seen in pregnancies complicated by medical conditions, predominantly diabetes and hypertension, with relatively higher cord HGF levels in these premature deliveries.

The higher levels of cord HGF in premature deliveries in complicated pregnancies could not be explained by one subset of medical conditions. One could speculate that there maybe a compensatory release of HGF from the placenta or fetal organs in these pregnancies secondary to intrauterine stress, but there was no inverse relationship between HGF and placental weight or birth weight to support this hypothesis. In support of the concept of compensatory release of HGF, mice homozygous for a targeted disruption of the HGF gene have severely impaired placenta development with a marked reduction in labyrinthine trophoblast cell numbers and consequently die in utero(23). A compensatory release of HGF by placental mesenchymal cells in a stressed placenta could help to preserve both placental function and weight.

Cord IGF-I levels in this study correlated with increasing birth size and gestational age. These observations have been made previously(15, 27) and act as further corroboration of the accuracy of gestational assessment. IGF-I levels tend to decrease with intrauterine growth retardation(15). IGF-I and HGF levels did not correlate making a direct effect of IGF-I on HGF release or action unlikely.

Previous evidence for HGF playing a role in human fetal growth and development is limited. Fetal rat liver, lung, and kidney express HGF(8), and HGF stimulates insulin production and islet cell cluster formation in cultured human fetal pancreatic cells, suggesting a role in islet development(28). The placenta is a rich source of HGF and immunohistochemical techniques demonstrate HGF in the placental mesenchyme and amniotic epithelium(7). The HGF receptor met is expressed by cytotrophoblasts(7).

Targeted disruption of the HGF gene in mice invariably results in death of the embryo due to a marked reduction of placental cytotrophoblastic cells(23, 24). Liver parenchymal mass is also markedly reduced with dissociated apoptotic liver parenchymal cells(24) and this defect may also render survival unlikely. These findings demonstrate that HGF is an essential factor for both placental growth and fetal organ growth and differentiation. Allelic disruption of the IGF-I gene results in profound growth retardation secondary to skeletal dwarfing and muscle hypoplasia(29, 30), but does not usually result in death.

HGF has no structural homologies with other known growth factors. It has the potential to act as a mitogen and motogen in vitro via direct interaction with its specific receptor encoded by the c-met oncogene(31). Its ability to induce endothelial cell motility in collagen gels by cell dissociation and migration, so called“scattering,” has been linked to angiogenesis and vascular repair(32). HGF stimulates directed and random migration of endothelial cells(5) and causes angiogenesis in vivo in the rabbit cornea(6). Met is strongly expressed by cells lining tubular structures(33). Renal epithelial cells form tubule like structures in vitro in response to HGF(20, 33, 34) and a similar response has been seen with lumen formation by human breast carcinoma cells and colon cancer cells(35). HGF antibodies inhibit the branching morphogenesis of the ureteric bud in primary organ culture(18). HGF stimulates the growth of fetal rat gastric and intestinal cells in vitro(19) and promotes the growth and differentiation of multipotent and erythroid hematopoietic progenitor cells(20, 22). HGF may also regulate primitive streak formation in the chick embryo(35). Therefore, there is accumulating evidence in vitro for the role of HGF in tissue morphogenesis. Our finding of a relationship between cord HGF and gestation rather than growth of the fetus and placenta in late normal pregnancy suggests that HGF may be a more significant factor in tissue and organ differentiation and development, than in linear fetal growth.