Main

HGF/SF is a disulfide-linked heterodimeric heparin-binding molecule secreted by embryonic mesenchymal cells and adult derivatives such as fibroblasts(1–3). It has roles in both normal embryonic development, oncogenesis, and in regeneration of adult organs(4). These actions are mediated by MET, a membrane-bound receptor tyrosine kinase which undergoes autophosphorylation upon binding HGF/SF and transduces growth signals into the cell(5, 6). In mice MET is expressed by diverse epithelia(1, 7, 8) and also by chondrocytes(9), endothelia(10), hematopoietic precursors(11), neurons(12), renal mesenchymal(7) and mesangial cells(13), and myogenic precursors(14).

Based on cell and organ culture experiments(1, 7, 15–17), as well as expression patterns of HGF/SF and MET in mice and rats(7, 18, 19), it was postulated that HGF/SF elicits embryonic epithelial growth and morphogenesis by paracrine or even autocrine signaling systems. Indeed, mice with homozygous null-mutations for either HGF/SF or met have impaired growth and die in utero due to perturbed trophoblast and hepatocyte differentiation(14, 20, 21).

Little is known, however, about the HGF/SF-MET axis in early human development. We now demonstrate that the ligand and receptor are widely expressed during critical early periods of human organogenesis between 6 and 13 wk after fertilization. In addition, the expression pattern of HGF/SF and MET by multicystic dysplastic kidney cells suggests a role for the ligand and receptor in a case of abnormal epithelial growth in a human organ malformation(22–24).

METHODS

Sources of tissues. Human fetuses were provided by the Medical Research Council-funded Human Embryo Bank at the Institute of Child Health, from collections made under local research ethical committee permission. Thirteen normal fetuses were collected from chemically induced and surgical terminations performed 5-13 wk after fertilization as described previously(25). Gestational age was determined by standard criteria from external morphology(26). All chemicals were supplied by Sigma Chemical Co. (Poole, Dorset, UK) unless otherwise stated. Embryos for histology were fixed in 4% paraformaldehyde in PBS (pH 7.0) overnight, embedded in paraffin wax, and sectioned at 5 μm. Other embryos were dissected in ice-cold Liebowitz-15 media (Life Technologies, Inc., Gaithersburg, MD), and organs were snap-frozen for biochemical analysis. A postnatal human dysplastic kidney harvested at nephrectomy was fixed, embedded in paraffin wax, and sectioned exactly as we have previously described(23, 24).

RT-PCR. RT-PCR was performed at each stage from two embryos. It should be emphasized that this RT-PCR methodology is sensitive and specific, but it is not quantitative. RNA was extracted from fetal organs using TRI-REAGENT (Molecular Research Center, Cincinnati, OH). Next, 300 ng of RNA were treated with deoxyribonuclease I to degrade contaminating DNA and subjected to RT-PCR as described elsewhere(7). Primers were: MET sense primer 5′-ACA GCG CGC CGT GAT GAA TAT CGA-3′, corresponding to nucleotides 1230-1254 of human MET cDNA (cDNA)(27); MET antisense 5′-GCA GCC CAA GCC ATT CAA TGG GAT-3′, corresponding to nucleotides 1536-1560 of human MET cDNA; HGF/SF sense 5′-TCC ATG ATA CCA CAC GAA CAC AGC-3′ corresponding to nucleotides 458-482 of human HGF/SF cDNA(28); and HGF/SF antisense 5′-TGC ACA GTA CTC CCA GCG GGT GTG-3′ corresponding to nucleotides 828-852 of human HGF/SF cDNA. Primers forβ-actin, a housekeeping gene, were obtained from Clontech (Palo Alto, CA). RNA was reverse transcribed and subjected to PCR as described elsewhere(7). The PCR machine (Quatro TC-40) was programmed for 30 cycles of 94 °C for 1 min; then 65 °C for 1 min for MET andβ-actin or 67 °C for 1 min for HGF/SF; then 72 °C for 1 min. Seven microliters of reaction products were electrophoresed through a 1% agarose gel with ethidium bromide. These protocols resulted in the following products: 330 bp for MET, 394 bp for HGF/SF, and 838 bp for β-actin. RT-PCR products were subcloned in Bluescript XL1 Blue (Stratagene, La Jolla, CA) and sequenced to confirm their identities (data not shown).

Immunohistochemistry. Sections were dewaxed through HistoClear(National Diagnostics, Atlanta, GA) followed by rehydration in 30-100% alcohol. After washing in PBS (pH 7.4) for 5 min, fetal sections were microwaved for 8 min. After cooling and washing in PBS, they were incubated in 3% hydrogen peroxide for 15 min to quench endogenous peroxidase activity. Nonspecific binding was blocked with 10% vol/vol FCS in PBS, and primary antibodies were applied for 1 h at 37 °C. Anti-HGF/SF raised in sheep against whole human factor (from E. Gherardi, ICRF, Cambridge, UK) and anti-MET rabbit antiserum raised against the carboxy terminus of human MET(Santa Cruz Biotechnology, Santa Cruz, CA) were used as described previously(13). Preliminary experiments showed no signal upon either omitting primary antibodies or preincubation at room temperature for 2 h with recombinant HGF/SF (a gift from E. Gherardi, ICRF, Cambridge, UK) or MET peptide (Santa Cruz Biotechnology); both were used at 10 times the concentration of primary antibodies(13). Primary antibodies were detected using a streptavidin biotin peroxidase system (ABC Kit, DAKO, Carpinteria, CA) followed by diaminobenzidine. Sections were counterstained with hematoxylin and mounted in dextropropoxyphene (BDH, Poole, UK) and examined on a Zeiss Axiophot microscope (Carl Zeiss, 7082 Oberkochen, Germany). Sections of human dysplastic kidney were treated similarly except that primary antibodies were detected with FITC-conjugated second antibodies as described previously(13). Slides were mounted in Citifluor (Chemical Labs, Canterbury, UK) and were examined under fluorescence(wavelength 488 nm) on a Leica confocal laser scanning microscope(Aristoplan-Leica, Heidelberg, Germany) as described previously(13).

Western blot. Organs were lysed in ice-cold radioimmune precipitation assay buffer (PBS with 1% Nonidet P-40, 10 mM sodium deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride, 200 nM aprotinin, and 1 mM sodium orthovanadate). Protein was measured by the BCA protein assay as described by the manufacturer (Pierce, UK), and 10 μg were used directly for HGF/SF Western blots or was first immunoprecipitated essentially as described previously(13) before HGF/SF or MET blots. Briefly, lysate was precleared with normal rabbit IgG and protein A-agarose(Santa Cruz Biotechnology). After centrifugation, the supernatant was reacted with anti-MET or anti-HGF/SF antibodies for 1 h at 4 °C, followed by addition of 20 μL of protein A-agarose and incubation overnight at 4°C. Proteins were resuspended and heat denatured at 100 °C for 5 min, separated by electrophoresis in a 0.2 M SDS-polyacryl-amide gel(29), and transferred onto nitrocellulose (Hybond-C Extra; Amersham, UK)(30). Membranes were incubated in blocking solution (5% nonfat milk in Tris-buffered saline) at 4 °C overnight and reacted with primary antibodies which were detected with horseradish peroxidase-conjugated second antibodies and enhanced chemiluminescence reagent (ECL, Amersham, UK). Proteins were sized with Rainbow markers (Amersham, UK).

RESULTS

HGF/SF and MET fetal mRNA. As assessed by RT-PCR, we found both genes expressed in the following organs between 6 and 13 wk of gestation: metanephric kidneys, adrenal glands, liver, lung, brain, stomach, intestines, heart, and placenta. We also found mRNA for HGF/SF and MET in mesonephric kidneys between 6 and 8 wk; these organs involute later in gestation. We also detected mRNA for both genes in skin and eye at 10 wk and in pancreas at 13 wk. In general there was coexpression of ligand and receptor mRNA, with the exception of spinal cord at 6 wk when only MET was detected. These results are summarized in Table 1, and typical RT-PCR products are shown in Figure 1 with each stage representing consistent data from two embryos.

Table 1 HGF/SF and MET mRNA as assessed by RT-PCR
Full size table
Figure 1
figure 1

RT-PCR of HGF/SF and MET in a 10-wk fetus. RT-PCR demonstrates HGF/SF and MET mRNA in kidney (metanephros), liver, lung, gut(intestine), and placenta of a 10-wk gestation human fetus. RT-PCR forβ-actin confirms mRNA integrity. No RT is a negative control(placental mRNA without reverse transcriptase). PCR amplification resulted in products: 394 bp for HGF/SF, 330 bp for MET, and 838 bp for β-actin.

Full size image

Overview of HGF/SF and MET fetal protein analysis. Many organs showed specific immunohistochemical signals for MET, as described below. In contrast, the placenta showed an intense immunohistochemical signal for HGF/SF, whereas no consistent specific staining pattern was observed in other organs. Because most organs expressed HGF/SF mRNA as assessed by the sensitive RT-PCR technique, we speculated that levels of HGF/SF protein were relatively low in nonplacental tissues. To investigate this further, equal amounts of protein from different organs (liver, lung, hands, feet, metanephric kidney, mesonephros, placenta, and gut) from a 10-wk embryo were studied by Western blots (Fig. 2). HGF/SF protein was detected in all specimens, but the signal was highest in the placenta.

Figure 2
figure 2

Western blot for HGF/SF in a 10 wk fetus. Ten micrograms of protein from organs were analyzed by Western blot for HGF/SF. Note a major immunoreactive band at approximately 60-65 kD, with less intense bands detected at approximately 28-30 and 90-97 kD. These species correspond to the α-chain, β-chain, and αβ-heterodimer HGF/SF precursor, respectively. The placenta appears to express the highest levels of HGF/SF; this is evident when comparing levels of the β-chain with other organs. The HGF/SF precursor appears to be prominent in liver and placenta. This is a representative Western blot from two separate experiments. Key:Liv, liver; Lun, lung; UL, upper limb;LL, lower limb; Meso, mesonephros; Met, metanephros; Pla, placenta; Int, intestine; and HGF, murine recombinant HGF/SF.

Full size image

Placenta. Chorionic villi are the fetal component of the placenta, and before 20 wk the placental barrier consists of a syncytiotrophoblast enveloping the cytotrophoblast layer, which itself encloses a core of connective tissue, Hofbauer phagocytic cells, and endothelia(26). Chorionic villi of a 5-wk human fetus are shown in Figure 3. The multinucleate syncytiotrophoblast and mononuclear cells of the cytotrophoblast stained intensely for HGF/SF protein. In contrast, MET immunostaining was confined to the cytotrophoblast. In addition, endothelia within the villus stained for both HGF/SF and MET, whereas a subset of connective tissue or phagocytic cells in the core of the villus stained for HGF/SF protein. Villi from later ages showed similar expression patterns. Finally, endothelia in the maternal component of the placenta, the decidua basalis, were positive for MET but not HGF/SF (data not shown).

Figure 3
figure 3

HGF/SF and MET in human trophoblast. (A) Chorionic villus of a 5-wk fetus stains intensely for HGF/SF. (B) Similar field as A after primary antibody was preabsorbed with HGF/SF. (C) Higher power demonstrates HGF/SF staining in syncytiotrophoblast (s) and cytotrophoblast (c). Note individual cells in the center of the villus also stain for HGF/SF(arrows). (D) MET immunostaining in chorionic villus.(E) Similar field as A after primary antibody was preabsorbed with MET peptide. (F) Higher power demonstrates that MET staining is prominent in cytotrophoblast (curved arrows), whereas surrounding syncytiotrophoblast (s) shows little staining. Note that capillaries in the core of the villus also express MET (arrowhead). Positive immunostaining appears brown, and nuclei are counterstained blue with hematoxylin. Bars are 80 μm in A, B, D, and E and 12μm in C and F.

Full size image

Liver. The liver arises as a ventral bud of the foregut endoderm in the 4th wk after fertilization and gives rise to the hepatocytes and biliary ducts. These epithelia interact with part of the splanchnic mesoderm called the septum transversum, which will contribute to the diaphragm and hematopoietic and Kupffer cells(26). We were unable to detect specific signals for either HGF/SF or MET using immunohistochemistry in the human liver between 6 and 13 wk, despite the presence of mRNA for both genes (Fig. 1 and Table 1). Therefore, we performed Western blots after immunoprecipitation of liver protein; appropriate bands for both HGF/SF and MET (Fig. 4) were detected at 10 wk of gestation and at earlier and later stages (data not shown).

Figure 4
figure 4

A 10-wk fetal liver Western blot after immunoprecipitation. (A) Blot of unreduced liver protein with anti-human HGF/SF polyclonal antibody; note the band at 92 kD representing the HGF/SF αβ-heterodimer (arrowhead). (B) Blot of reduced liver protein with polyclonal antibody raised against whole human HGF/SF. Note bands at 60 and 30 kD representing the α- and β- chains of HGF/SF (arrowheads). (C) Blot of reduced liver protein with antibody raised against the MET carboxy terminus. Note the band at 140 kD representing the MET β-chain (arrowhead). When respective antibodies were reacted with recombinant HGF/SF or MET peptides before application to the Western blots, no signals were detected (data not shown). Note that the longitudinal scale of each blot is different and that sizes of Rainbow markers are also indicated.

Full size image

Kidney. The metanephroi are direct precursors of the adult kidneys(26). They are first detected at 6 wk when the ureteric bud interacts with intermediate mesoderm called renal mesenchyme; the first glomeruli form by 9 wk. At this stage an intense signal for MET was detected in ureteric bud branches (Fig. 5,A and D) and in immature nephrons, the S-shaped bodies (Fig. 5A). In these epithelia, immunostaining was localized to the basal plasma membrane. A lower intensity of MET staining was observed in condensates and vesicles, which are cells undergoing a conversion from a mesenchymal to epithelial phenotype (Fig. 5A). In fetal glomeruli, positive MET immunostaining was found in Bowman's capsule and in endothelial and mesangial cells situated in the core of the glomerulus (Fig. 5C).

Figure 5
figure 5

MET immunohistochemistry in a 9-wk metanephros.(A) Positive MET staining in ureteric bud branches (u) and primitive nephrons such as vesicles (v) and S-shaped bodies(s). Note marked staining on basal surface of primitive epithelia(arrows). Condensing renal mesenchyme (m) expresses a much lower level of MET in the cytoplasm. (B) Similar field as A after primary antibody was preabsorbed with MET peptide; no specific staining. (C) MET staining in basal surface of epithelia of Bowman's capsule (open arrows) of fetal glomeruli (g). Mesangial (arrow) and endothelial cells in the fetal glomeruli are MET-positive. (D) The stalk of the ureteric bud shows positive MET immunostaining on the basal epithelial surface. MET immunostaining appears brown, and nuclei are counterstained blue with hematoxylin. Bars are 20μm.

Full size image

Other sites of MET expression. The lung bud develops at 3-4 wk and subsequently branches in the surrounding splanchnic mesenchyme. Between 5 and 13 wk, the lung is pseudoglandular, resembling an exocrine gland(26). Figure 6A shows a 13-wk organ where MET immunostaining was observed in the apical epithelial domain. Cartilaginous plates develop from the splanchnic mesenchyme around the tracheal and large bronchial epithelia; here we observed MET staining in chondrocytes (Fig. 6C). MET immunostaining was detected in the basal domain of fetal intestinal epithelia (Fig. 6E) and in perineural tissue around dorsal spinal nerves (Fig. 6G). Endothelia stained for MET in many organs including placenta (Fig. 3F), kidney (Fig. 5C), and heart (not shown). MET immunostaining was also detected in cartilage of the appendicular skeletal (fetal digits) and vertebrae of the axial skeleton (data not shown), both of which undergo endochondral ossification.

Figure 6
figure 6

MET expression in fetal lung, intestine, and spinal nerve root. (A) A 13-wk fetal lung showing apical MET staining(arrowheads) in the primitive branching epithelium. (B) Similar field as A after primary antibody was preabsorbed with MET peptide. (C) A 13-wk fetal trachea with marked MET expression by chondrocytes. (D) Similar field as C after primary antibody was preabsorbed with MET peptide. (E) A 13-wk fetal intestine with MET staining on the basal surface (arrowheads) of the epithelial cells (e). (F) Same field as E after primary antibody was preabsorbed with MET peptide. (G) Dorsal root nerve from a 9-wk fetus shows MET staining in perineural tissue. (H) Similar field as G after primary antibody was preabsorbed with MET peptide. MET immunostaining appears brown, and nuclei are counterstained blue with hematoxylin. Bars are 80 μm in C-F and 12 μm in the other frames.

Full size image

HGF/SF and MET in a kidney malformation. Multicystic dysplastic kidneys contain dysplastic tubules which terminate in cysts; these structures are considered to be aberrant branches of the ureteric bud(22). HGF/SF immunolocalized to stromal and mesenchymal cells surrounding dysplastic tubules, and also to dysplastic epithelia themselves (Fig. 7A). HGF/SF immunoreactivity was also noted on the epithelial surface of cysts and in amorphous material in cyst lumens (Fig. 7C). MET protein was detected in the apical, lateral, and basal membrane domains of dysplastic tubule and cystic epithelia (Fig. 7,C and D). Stromal and undifferentiated cells did not express MET, as assessed by this technique.

Figure 7
figure 7

HGF/SF and MET immunohistochemistry in a multicystic dysplastic kidney. (A) A dysplastic tubule within a postnatal human kidney malformation. Note HGF/SF immunostaining localized to the epithelium itself and also to scattered cells in the surrounding undifferentiated stromal/mesenchymal tissues. (B) In a section adjacent to that depicted in A, MET immunostaining was confined to the epithelium. dt, dysplastic tubule. (C) Other parts of the same organ contained large cysts. Faint HGF/SF immunoreactivity was noted in three locations: stromal around the cyst, apparently coating the lining of the cyst, and in amorphous material inside the cyst. (D) In a section adjacent to that depicted in C, MET immunostaining was prominent in epithelium lining the cyst (cy). Bar is 50 μm.

Full size image

DISCUSSION

Our study shows that HGF/SF and MET mRNAs are widely expressed in the human embryo during early organogenesis from 6 wk of gestation when differentiation often proceeds by mesenchymal/epithelial interactions(26). Hence HGF/SF may be an important growth factor in early human embryogenesis. These results extend observations from a study which used Northern blots to detect HGF/SF and MET mRNA and in situ hybridization to localize HGF/SF transcripts in fetuses between 9 and 17 wk(31). That report showed HGF/SF and MET mRNAs were expressed by a similar range of organs as we have studied. Some human organs such as the metanephros, however, are relatively differentiated by 9 wk, and therefore the earlier time-frame of our study may be more relevant to early organogenesis(26). Furthermore, Wang et al.(31) did not assess the localization of MET protein or measure HGF/SF protein.

We found organs with a very specific pattern of MET immunostaining. In fetal kidney, gut, and lung the expression of the HGF/SF receptor was marked in epithelia, a pattern consistent with a studies in embryonic mice(18). MET protein was located in basal plasma membranes of metanephric and gut epithelia, where it would be well placed to receive a paracrine signal from HGF/SF secreted by adjacent mesenchyme(18). In fact, functional animal experiments implicate mesenchyme-derived HGF/SF in epithelial growth during nephrogenesis(7) and in the adult gut(8). In contrast, we found that MET immunostaining was localized apically in the branching epithelia of the developing lung, in which case the ligand would need to access the luminal surface. HGF/SF null-mutant mice have abnormal liver development(20), and HGF/SF stimulates DNA synthesis in fetal rat hepatocytes(19). Our study confirms that MET mRNA and protein were present in very early human hepatogenesis, although we were unable to immunolocalize the protein.

We found that MET expression was not confined to epithelial cells. Fetal chondrocytes in bronchi as well as in axial and appendicular skeleton expressed MET, a finding consistent with an animal study in which HGF/SF stimulated proliferation and proteoglycan synthesis in adult chondrocytes(9). Similarly, we found low levels of MET in renal mesenchymal condensates and fibromuscular mesangial cells, consistent with our previous studies of developing mice(7, 13). Human fetal endothelia also expressed MET, and there is animal evidence that HGF/SF can act as an angiogenic factor(10).

We found that many developing tissues expressed HGF/SF mRNA as assessed by RT-PCR. However, it was not possible to obtain a specific immunohistochemical signal for HGF/SF in these tissues, with the exception of chorionic villi. HGF/SF protein was detected by Western blot in a variety of fetal tissues with the strongest signal in the placenta. We therefore speculate that levels of HGF/SF protein are relatively low in nonplacental tissues. If secreted HGF/SF were to have the ability to diffuse widely, then this would provide another explanation for our failure to detect specific HGF/SF immunohistochemical signals in most organs. Indeed, HGF/SF circulates in the human fetal plasma, and levels increase in late gestation(32). The sources of fetal plasma HGF/SF are undefined but might include placenta, the fetus itself, or transmission from maternal circulation.

Our Western blot data suggest that the placenta is a rich source of HGF/SF protein, and this organ is known to be a rich source of other growth factors(33). We found that both the syncytio- and cytotrophoblast contain high levels of HGF/SF protein, confirming a recent study of Wolf et al.(33). Another report, using in situ hybridization, demonstrated that the syncytiotrophoblast expresses HGF/SF mRNA(31). Of note, HGF/SF stimulates the proliferation of cytotrophoblasts in vitro(34), consistent with our observation that MET immunostaining was prominent in these cells in vivo. Although classical studies have emphasized that HGF/SF expression is generally a property of mesenchymal cells(1), Wang et al.(31) found that epithelia of the human fetal gut, tongue, and skin expressed this gene using in situ hybridization. Although these data appear to contradict expression patterns of HGF/SF in mouse development(18), if embryonic HGF/SF is indeed made by some epithelial as well as mesenchymal cells, there would be scope for autocrine as well as paracrine modes of action. We have tried to perform in situ hybridization for HGF/SF in human embryos; thus far we have yet to obtain a convincing signal from any tissue apart from placenta (M.K.-J. and A.S.W., unpublished observations).

Recently, a human cranial malformation (Crouzon syndrome) was linked to mutation of the fibroblast growth factor receptor 2 gene(35). Like MET, this is a receptor tyrosine kinase. Human fetal deaths or organ malformations have yet to be linked to mutations of either HGF/SF or MET, both located on chromosome 7. However, we considered it instructive to examine the expression of this signaling system in the multicystic dysplastic kidney, an example of abnormal epithelial/mesenchymal interaction. The etiology of these malformations is unknown(22), but they contain tubules and cysts, which are abnormal collecting ducts, surrounded by stromal cells, considered to be mesenchymal cells that fail to form nephrons(22–24). We previously demonstrated that dysplastic epithelia express PAX2, a potentially oncogenic transcription factor, and maintain a high rate of proliferation postnatally(24). Other data from our laboratory, using mouse metanephros as a model, showed that renal mesenchyme secretes bioactive HGF/SF; moreover, blocking antisera to HGF/SF prevented the growth of primitive collecting ducts in organ culture(7). Another study, using similar methodologies, concluded that HGF/SF enhanced elongation of ureteric bud branches(36).

Our preliminary observations show that dysplastic renal tubules and cysts express MET, and furthermore, HGF/SF immunolocalizes to dysplastic tubules and surrounding stromal cells (Fig. 7). Another laboratory reported that MET immunostaining was barely detectable in collecting ducts of normal postnatal kidneys(37), and our own unpublished data demonstrate that HGF/SF and MET immunoreactivity is minimal in the mature kidney (A.S.W. and P.J.D.W., personal observations). Therefore, our descriptive observations suggest that HGF/SF may play a role in growth of dysplastic epithelia. Two other studies are of relevance here. Natali et al.(37) found that MET was overexpressed in human renal carcinomas, whereas Horie et al.(38) reported that HGF/SF and MET mRNA were coexpressed in walls of hyperproliferative autosomal dominant polycystic kidney disease cysts; moreover, immunoreactive HGF/SF was detected in cyst lumens. At present we are constructing cell lines from dysplastic kidneys to investigate functional aspects of HGF/SF and MET in vitro. In future it will be interesting to study the expression of HGF/SF and MET in malformations of other human organs.