Activation of hepatocyte growth factor (HGF) by endogenous HGF activator is required for metanephric kidney morphogenesis in vitro.

The interaction of hepatocyte growth factor (HGF) with c-Met has been implicated in morphogenesis of the kidney, lung, mammary gland, liver, placenta, and limb bud. HGF is secreted as an inactive zymogen and must be cleaved by a serine protease to initiate Met signaling. We show here that a serine protease specific for HGF, HGF activator (HGFA), is expressed and activated by the ureteric bud of the developing kidney in vivo and in vitro. Inhibition of HGFA activity with serine protease inhibitors reduced ureteric bud branching and inhibited glomerulogenesis and nephrogenesis. Activated HGF rescued developing kidneys from the effects of inhibitors. HGFA was localized around the tips of the ureteric bud in developing kidneys, while HGF was expressed diffusely throughout the mesenchyme. These data show that expression of HGF is not sufficient for development, but that its activation is also required. The localization of HGFA to the ureteric bud and the mesenchyme immediately adjacent to it suggests that HGFA creates a gradient of HGF activity in the developing kidney. The creation and shape of gradients of activated HGF by the localized secretion of HGF activators could play an important role in pattern formation by HGF responsive tissues.

The interaction of hepatocyte growth factor (HGF) 1 with the receptor tyrosine kinase c-Met is a prototype of mesenchymal to epithelial signaling. HGF has mitogenic, motogenic, and morphogenic activities in a variety of cell types (1)(2)(3)(4). It has been implicated in mesenchymal to epithelial signaling in the development of the kidney, lung, mammary gland, liver, and placenta (5). In the kidney, HGF is produced solely by the metanephric mesenchyme (6,7). The Met receptor is expressed by metanephric mesenchyme as well as by the epithelium of the ureteric bud and later by S-shaped bodies and tubules. Neutralizing antibodies against HGF blocked branching morphogenesis by the ureteric bud in organ cultures of kidney rudiments and inhibited the early steps in branching morphogenesis by immortalized ureteric bud cells (UB cells) in threedimensional organ culture (7,8). Antibodies against HGF, in addition to their effects on branching morphogenesis, inhibited glomerulogenesis and nephrogenesis.
HGF is a member of the plasminogen family of serine proteases although it lacks proteolytic activity. Like other members of this group, it is secreted as an inactive zymogen that must be activated by proteolytic cleavage at Arg 494 -Val 495 (9 -12). A specific serine protease, HGF activator (HGFA), was purified from serum based on this biological activity and has been shown to mediate activation of HGF in injured tissues in vivo (13,14). Other serine proteases, including urokinase, factor XIIa, and tissue plasminogen activator have been shown to activate HGF in vitro, but their biologic role in HGF activation is unknown (11,15,16). HGFA is expressed in adult liver and is secreted into the blood (14). Like other serine proteases in the coagulation and fibrinolytic cascades, HGF activator is secreted as an inactive single chain zymogen. It is cleaved to the catalytically active, disulfide-linked heterodimer by thrombin in vitro. The expression of HGFA during embryonic development is not known. HGF is required for embryonic development (17,18), so that activators of HGF must be expressed during embryogenesis. What might be the source of HGF activators in the developing metanephric kidney? The organ culture experiments that revealed a role for HGF in renal development contained no serum and hence no exogenous HGF activating activity. Therefore, the developing kidney must contain an endogenous HGF activator. A likely candidate for this activity is HGFA.
To test the hypothesis that HGFA is responsible for HGF activation during fetal development, we cloned the murine homolog of the human HGFA gene. In the earliest stage of renal development, message for HGFA was found only in microdissected ureteric bud and was not detected in mRNA extracted from the adjacent mesenchyme. Immortalized UB cells expressed HGFA message, protein, and an HGF hydrolyzing activity that was inhibited by a variety of serine protease inhibitors including leupeptin. UB cells scattered in response to the single chain zymogen of HGF; this scattering activity was inhibited by the same concentrations of leupeptin that blocked HGF activation. Leupeptin at concentrations that blocked the hydrolysis of HGF by ureteric bud in vitro inhibited renal morphogenesis in organ culture. The inhibitory effect of * This work was supported in part by March of Dimes Grant 1-FY98-0513. Confocal microscopy was performed at the Confocal Microscopy Facility of Columbia University, which was established by National Institutes of Health Shared Instrumentation Grant 1S10 RR10406 and is supported by National Institutes of Health Grant 5-P30-CA13696 as part of the Herbert Irving Cancer Center at Columbia University. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF224724.
‡ Established Investigator of the American Heart Association. To whom correspondence should be addressed. Tel.: 212-305-4476; Fax: 212-305-3475; E-mail: jsv1@columbia.edu. § Partially supported by the Summer Undergraduate Research Fund project of Columbia University. 1 The abbreviations used are: HGF, hepatocyte growth factor; UB cell, ureteric bud cell; HGFA, hepatocyte growth factor activator; EST, expressed tag sequence; RT-PCR, reverse transcriptase-polymerase chain reaction; PAGE, polyacrylamide gel electrophoresis; PBS, phosphate-buffered saline; kb, kilobase(s). leupeptin could be rescued by activated HGF. These data show that the activation of HGF in developing kidney is kidney autonomous and suggest that activation of HGF is regulated by ureteric bud secretion of HGFA.

MATERIALS AND METHODS
Sequence of Murine HGFA-The coding sequence of human HGFA (GenBank TM accession number D50030) was used to screen the murine EST data base using the BLAST algorithm (19). Sequences identified by the initial screen were used to re-interrogate the data base. By this process, 9 clones were identified and assembled into a contig. Five of these clones (mx85 g08, uc76a08, mj60f03, mw81a02, and ma52a08) were purchased from ATCC, characterized by restriction mapping, and sequenced in both directions. The sequence of murine HGFA was deposited in GenBank TM (AF224724).
Northern Blotting-A 942-nucleotide EcoRI, NarI fragment of IMAGE clone mj60f03 corresponding to nucleotides 922 to 1864 of the full-length sequence was used as a probe. The probe was labeled with digoxigenin-UTP by random priming and used to screen mouse multiple tissue Northern blots (CLONTECH). For Northern blots of UB cell RNA, total RNA was isolated from UB cell cultures using RNAzol (Cinna) according to the manufacturer's instructions. Total RNA (1 g) was denatured and transferred to nylon membranes (Immobilon-N, Millipore). The membrane was hybridized for 16 h at 45°C and washed with 0.1 ϫ SSC containing 1% SDS at 65°C. Bound probe was visualized with alkaline phosphatase and chemiluminescent detection system according to the manufacturer's instructions (Roche Molecular Biochemicals).
RT-PCR-RNA was isolated from murine UB cells or from microdissected rat ureteric buds and mesenchymes using RNAzol (Cinna). Poly(A) RNA from rat ureteric buds and mesenchymes was purified from total RNA by affinity chromatography on poly(dT) resin (Oligotex-dT, Qiagen). The microdissected rat kidney rudiments were obtained at embryonic day 13.5, when the ureteric bud had branched a single time. This stage of kidney development in the rat is morphologically equivalent to embryonic day 11.5 in murine kidney development.
Primers for RT-PCR of murine HGFA were TGTTCAACCCCAA-CAACC (forward primer, nucleotides 1474 -1491) and GTAAGCCA-CACCATTCTTCTC (reverse primer, nucleotides 1838 -1818). Total RNA was transcribed into cDNA using avian myeloblastosis virus reverse transcriptase and random hexamers. The cDNA was subjected to 30 cycles of PCR using Taq polymerase (Promega). The annealing temperature was 59°C for PCR with the murine template (UB cell RNA) and 54°C for PCR with the rat template (dissected ureteric bud or mesenchyme RNA). Amplified DNA fragments were purified from agarose gels using QIAEX resin (Qiagen) and directly sequenced by the DNA facility of Columbia University.
Cells, Plasma, Serum, Conditioned Medium, and HGF-UB cells (20) were grown in Dulbecco's modified Eagle's medium/F-12 (Life Technologies, Inc.) with 5% fetal bovine serum (Hyclone) and 1% ITS ϩ supplement (insulin, transferrin, selenous acid, essential fatty acids), from Fisher Scientific. Conditioned medium was obtained by washing confluent cultures of UB cells three times with Dulbecco's modified Eagle's medium/F-12 containing no additives. The cells were incubated for 8 h at 37°C, and washed again three times. Twenty-five ml of medium without additives were then incubated with confluent cultures in 150cm 2 flasks for 48 h to obtain conditioned medium. The conditioned medium was cleared by centrifugation at 3000 ϫ g for 15 min and concentrated 1000-fold with a Centricon Plus-80 Biomax-8 device (molecular mass cutoff 8000 daltons, Millipore). Mouse plasma and serum were purchased from Pelfreez. Activated HGF was obtained from R & D Systems (NSO HGF). The inactive zymogen of HGF (HGFsc) was a gracious gift of Dr. George vande Woude (National Institutes of Health, Bethesda, MD).
Antibodies-Antibodies raised against peptides derived from human HGFA were purchased from Santa Cruz Biotechnology. The peptide used to raise antibody HGFA-L (N-19) is identical in murine and human HGFA (underlined in Fig. 1). Antibodies against human HGF ␣ subunit (N-19 and C-20), WT1 (C- 19), and the HGF receptor met were also obtained from Santa Cruz Biotechnology. Antibody against phosphotyrosine (4G10) was obtained from Upstate Biotechnologies. TROMA-1 was obtained from the Developmental Studies Hybridoma Bank. Cy3labeled anti-rabbit and fluorescein isothiocyanate-labeled anti-rat IgG were obtained from Jackson ImmunoResearch. Alexa 488-labeled antirabbit antibody was obtained from Molecular Probes. For some experiments, TROMA-1 antibodies were purified by anion exchange chromatography and labeled with Alexa 488 (Molecular Probes) according to the manufacturer's instructions.
Immunoblots-Proteins were subjected to SDS-PAGE under reducing conditions and transferred to polyvinylidene difluoride (Immobilon-P, Millipore). Blots were blocked for 1 h with 10% instant nonfat dry milk in PBS and then incubated overnight at 4°C in primary antibody diluted in 1% bovine serum albumin. Concentrations of primary antibody were 1/100 for anti-HGF and 1/1000 for anti-phosphotyrosine and anti-Met. Blots were washed 3 times in 5% milk and incubated for 45 min in 1:10,000 dilution of peroxidase-labeled secondary antibody (Jackson ImmunoResearch). Antibody binding was visualized by chemiluminescence (ECL, Amersham Pharmacia Biotech).
Assay for HGFA Activity-Serial dilutions of UB conditioned medium were incubated for 2 h at 37°C with 200 g/ml single chain HGF (HGFsc). Aliquots of the mixture were removed at various times and analyzed by SDS-PAGE under reducing conditions.
Immunoprecipitation of Met-Confluent cultures of UB cells were incubated overnight in serum-free medium. The next day, HGFsc was added to the culture medium in the presence or absence of leupeptin or serum and the cultures were incubated for 15 min at 37°C. The cultures were then washed three times in PBS and scraped in SB (150 mM NaCl, 10 mM Tris-HCl, pH 7.4, 10 mM EDTA, Complete protease inhibitors from Roche Molecular Biochemicals, 1 mM sodium orthovanadate, and 1% Triton X-100). The extracts were incubated for 30 min at 4°C and pelleted for 15 min at 10,000 ϫ g to obtain cleared lysates. Lysates were incubated for 1 h with 1 g of rabbit anti-Met and then for a second hour with Protein A-Sepharose (Roche Molecular Biochemicals). The precipitates were washed three times with SB and displayed on a 5% reducing SDS-PAGE. The immunoblots were processed for with anti-phosphotyrosine as described above except that the blots were blocked in 1% bovine serum albumin in Tris-buffered saline (150 mM NaCl, 10 mM Tris-HCl, pH 7.4). After visualization of phosphotyrosine, the blots were stripped with 25 mM glycine, pH 2.5, and 1% SDS and then re-probed with anti-Met as described above.
Organ Culture of Fetal Kidney Rudiments-Kidney rudiments were dissected from timed pregnant Swiss Webster mice at E11.5 when the ureteric bud had branched once. Organs were explanted onto permeable supports (Transwell Clear, 3.0-m pore size, Costar) in chambers containing 0.25 ml of medium. Cultures were maintained at the air-water interface in 5% CO 2 for 96 h at 37°C and then fixed for 10 min in methanol at Ϫ20°C. Kidney rudiments were grown in Dulbecco's modified Eagle's medium/F-12 containing 1% ITS ϩ (Collaborative Research) and no serum. For HGF and HGFA staining, explants of freshly dissected E12.5 kidneys were incubated for 3-4 h on Transwells and then processed for confocal microscopy.
Confocal Microscopy of Fetal Kidney Rudiments and Image Analysis-Methanol-fixed organ cultures were stained en bloc as described previously (21). Briefly, the fixed organ cultures were re-hydrated in PBS and blocked for 6 h in 1% bovine serum albumin or 10% donkey serum (Jackson ImmunoResearch). The cultures were incubated in primary antibody for 24 -48 h at 4°C, washed for 8 -24 h in 1% bovine serum albumin, and incubated overnight in secondary antibody. The cultures were washed for 2-24 h, post-fixed for 10 min with 2% paraformaldehyde, quenched with 50 mM NH 4  Whole mounts were analyzed by confocal microscopy on a Zeiss 410 laser scanning confocal microscope equipped with an argon-krypton laser. The kidney rudiments were visualized with a ϫ 10 or 40 objective on a Zeiss Axiovert 100 using the 568-and 488-nm laser emission peaks for excitation. Images were collected using a 515-540-nm band pass and a 590-nm long pass filter set. Each kidney rudiment was scanned 10 times 5-m apart. The scans were recorded as TIFF files and used for image analysis as described previously (21).

RESULTS
Sequence of Murine HGFA-Interrogation of the data base of murine ESTs with the coding sequence of human HGFA yielded 9 clones, of which one (mx85g08) contained the complete coding sequence of murine HGFA. One clone (uc76a08) had a 47-nucleotide out of frame deletion in the 5Ј end. Clone mj60f03 contained the signal sequence which was followed by a 666-nucleotide in-frame deletion.
The sequence of murine HGFA was 82% identical to that of human HGFA at both the nucleotide and amino acid levels ( Fig. 1). The catalytic domains of the two proteins were 93% identical, with 18 conservative amino acid substitutions and a single nonconservative substitution (Leu 553 in mouse for Arg 555 in human) in the COOH-terminal 257 amino acids. The thrombin cleavage site at Arg 405 -Ile 406 that is required for activation of HGFA was conserved, as were all the amino acids required for catalytic activity. The fibronectin, epidermal growth factor, and Kringle domains were conserved as were the critical cysteines of all these motifs. The 104 amino acids at the aminoterminal end of the sequence were the least conserved, with 42 amino acid substitutions and 3 amino acid deletions in the murine relative to the human sequence.
We used the murine HGFA sequence to search the Gen-Bank TM data base with the BLAST algorithm and identified only human HGFA and a partial rat HGFA sequence. A human protein similar to HGFA has recently been identified (22,23) (GenBank TM accession number AAB46909). This molecule, hyaluronin-binding protein, was only 38% identical to murine HGFA at the amino acid level. As our murine HGFA sequence is 82% identical to the human HGFA sequence, our sequence is murine HGFA, not the murine hyaluronin-binding protein.
Murine HGFA Was Detected Only in Adult Mouse Liver by Northern Blotting-We probed Northern blots of mRNA isolated from fetal mice and a variety of adult mouse organs and found that HGFA was expressed only in adult mouse liver (Fig.  2, A). This result is consistent with the previous finding that human HGFA was expressed only in adult liver by Northern blotting (14). A single transcript of 2.0 kb was detected in mouse liver, which contrasts with the situation in man, where a minor transcript of 3.4 kb was also detected. No HGFA was detected in Northern blots of total embryo poly(A) RNA. These data suggested that either HGFA is not expressed in the developing animal or that it is expressed in concentrations too low for detection by Northern blotting. In support of the latter hypothesis, we found that 7 of the 9 ESTs for murine HGFA were isolated from embryonic mouse libraries. We also identified human ESTs for HGFA in adult ovarian tumor (T72625), fetal liver/spleen (R89811), and a pooled adult lung/testis/B cell library (AI243638 and AA897612), suggesting that HGFA is expressed at low levels in a variety of organs during fetal and adult life in man.
Message for HGFA Is Expressed Exclusively in the Ureteric Bud in Early Renal Development-To test the hypothesis that HGFA is expressed during kidney development, we screened poly(A) RNA isolated from microdissected rat ureteric buds and mesenchymes by RT-PCR. We found that the predicted 345base pair fragment was consistently amplified from the ureteric bud but not from the mesenchyme (Fig. 2, C). Treatment of RNA with DNase prior to reverse transcription and PCR did not abolish amplification of this fragment. However, treatment of the samples with RNase blocked amplification. The amplified fragment was purified and sequenced. The sequence was similar, but not identical, to the sequence of murine HGFA. In the area where the fragment overlapped with a partial rat HGFA sequence (GI:3116329), the sequence was identical to the rat sequence. Thus, the amplified fragment arose from rat RNA and could not have been derived from contamination by murine HGFA plasmid DNA. These data show that, during early renal development, HGFA is expressed exclusively by the ureteric bud.
Immortalized Ureteric Bud Epithelial Cells Express HGFA Message and Protein-To further characterize HGFA in ureteric bud epithelial cells, we screened an immortalized ureteric bud cell line (UB cells) for HGFA expression. The predicted 345-base pair fragment was amplified from UB cell RNA by RT-PCR (Fig. 2, C, UB cells). The sequence of this fragment was identical to the murine HGFA sequence. A single transcript of 2.0 kb was identified in UB RNA by Northern blotting using our murine HGFA probe (Fig. 2, B). These data support the results of the RT-PCR analysis of isolated ureteric buds and mesenchymes and show that ureteric bud epithelial cells express HGFA both in vivo and in vitro.
To determine whether UB cells synthesized and secreted HGFA protein, we probed immunoblots of conditioned medium from UB cells with antibodies to HGFA. Antibody HGFA-L (N-19) was raised against a peptide at the amino-terminal end of the 34-kDa catalytic subunit of HGFA (underlined in Fig. 1). It recognized polypeptides of 99 and 34 kDa in murine plasma, representing the intact, uncleaved high molecular mass zymogen of HGFA (99 kDa) and the 34-kDa catalytically active form (Fig. 3A). Only the 34-kDa catalytically active form was de-tected in serum (Fig. 3A) and in UB cell conditioned medium (Fig. 3B), suggesting that UB cells are capable not only of secreting but also of activating HGFA.
Localization of HGF and HGFA Proteins in Early Renal Development-We examined the distribution of HGFA in early kidney rudiments by confocal microscopy of whole mounts of freshly isolated E12.5 kidneys. We found that HGFA protein was expressed in the ureteric bud, as shown by the orange staining indicating colocalization of HGFA (green) and the ureteric bud marker (TROMA-1, red, arrowhead in Fig. 4A) (24). In the mesenchyme, HGFA was localized in a reticular pattern extending 2 to 3 cell diameters from the ureteric bud surface, consistent with deposition of HGFA, a heparin-binding protein (13), in the extracellular matrix surrounding the mesenchymal cells adjacent to the ureteric bud (arrows, Fig. 4A). Staining was specific for HGFA, as co-incubation of the antibody with the immunizing peptide blocked HGFA labeling but not labeling with the antibody for the ureteric bud specific cytokeratin, TROMA-1 (Fig. 4B).
Single optical sections suggested that HGFA was concentrated around the ureteric bud tips (Fig. 4A). To test this hypothesis, serial optical sections of a kidney rudiment stained for HGFA were optically summed and projected onto a single plane. These images were displayed with a pseudocolor algorithm to show the relative intensity of HGFA staining throughout the rudiment (Fig. 4C). HGFA immunoreactivity was concentrated around the ampulla of the ureteric bud, shown outlined in white. The distribution of HGFA was different from the distribution of HGF (Fig. 4D). HGF was localized throughout the mesenchyme and was not concentrated around the ureteric bud ampulla. This result shows that the high concentration of HGFA around the ureteric bud is not an artifact of the high local concentration of mesenchymal cells around the ureteric bud.
Conditioned Medium from UB Cells Contains HGFA Activity-The identification of a 34-kDa form of HGFA in UB conditioned medium suggested that UB cells secrete and activate HGFA. To test this hypothesis, we incubated the uncleaved zymogen of HGF (HGFsc) with serial dilutions of conditioned medium. HGFsc was completely converted to the active heterodimer (HGF␣ and HGF␤) by 2.4 g of protein from UB conditioned medium (Fig. 5A). These data show that UB cells secrete HGFA and hydrolyze HGFsc to a disulfide-linked heterodimer. The molecular weights of the HGF subunits produced by UB conditioned medium are consistent with activation of HGF by hydrolysis between Arg 494 and Val 495 .
HGFA is a serine protease whose HGF hydrolyzing activity is blocked by the serine protease inhibitor aprotinin and the serine and cysteine protease inhibitor leupeptin, but not by the acid protease inhibitor pepstatin or the aminopeptidase inhibitor bestatin (10,11,15). HGF hydrolysis by 2.5 g of protein from UB conditioned medium was inhibited by leupeptin and aprotinin (Table I and Fig. 5B). Bestatin, EDTA, phosphoramidon, E64, and pepstatin A did not block HGF hydrolysis by UB conditioned medium.
Activated HGF binds to the Met receptor tyrosine kinase and increases cell motility, producing scattering of islands of HGFresponsive cells in two-dimensional culture. To discover whether the HGF hydrolyzing enzyme secreted by UB cells produced functionally active, heterodimeric HGF from the inactive zymogen (HGFsc), we incubated UB cells with HGFsc in the presence or absence of leupeptin. In serum-free medium without HGF, UB cells were closely associated in islands with smooth borders (Fig. 6, A, no HGF). Activated, heterodimeric HGF produced scattering (Fig. 6, A, activated HGF) as did an equal concentration of HGFsc (Fig. 6, scHGF). The scattering  4. Localization of HGFA and HGF in developing kidneys by confocal microscopy. All kidney rudiments were isolated at E12.5 and were stained with TROMA-1 (red) to localize the ureteric bud. A and B, whole mount staining for HGFA in the absence (A) or presence (B) of a 10-fold molar excess of peptide immunogen. Both photomicrographs are 1-m optical sections. A, HGFA was found in the extracellular matrix surrounding mesenchymal cells (arrows). Staining was particularly intense within two to three cell diameters from the ureteric bud. The yellow-orange color of the T-shaped ureteric bud (arrowhead) shows that HGFA colocalized with the ureteric bud marker, TROMA-1, consistent with its expression in this tissue. B, peptide competition of HGFA staining, using a 10-fold molar excess of peptide immunogen. The peptide blocked staining with antibody to HGFA (green) but did not affect staining with TROMA-1 (red). Note that TROMA-1 is red rather than the yellow-orange color representing colocalization as seen in A. C and D, projections of 25 optical sections of kidney. The images were color coded for intensity of fluorescence, where blue is the lowest intensity and red is the highest intensity. Outline of the ureteric bud is shown in white for reference. C, HGFA was concentrated around the T shaped ampulla of the ureteric bud. D, HGF was found diffusely throughout the kidney. Scale bar in C is 50 m.
activity of HGFsc was inhibited by 100 and 10 M leupeptin (Fig. 6A, 100 M leupeptin and 10 M leupeptin. Note that the UB cells were viable after incubation even in high concentrations of leupeptin (100 M, Fig. 6A).) These data show that UB cells responded to HGFsc in a leupeptin-dependent manner.
To demonstrate further that UB cells produced the activated form of HGF from the inactive zymogen HGFsc, the HGF receptor met was immunoprecipitated from cultures of UB cells (Fig. 6B). The Met receptor immunoprecipitated from UB cells incubated overnight in serum-free medium contained little phosphotyrosine (-HGFsc-serum-leupeptin). HGFsc incubated with fresh medium containing serum for 15 min stimulated Met autophosphorylation, as predicted from the observation that serum contains HGFA (ϩHGFsc ϩserum-leupeptin). HGF zymogen incubated with UB cells incubated in serum-free medium overnight also stimulated Met autophosphorylation, showing that UB cells were able to activate and respond to HGFsc in the absence of exogenous HGFA from serum (ϩHG-Fsc-serum-leupeptin). Activation of HGF signaling was substantially inhibited by leupeptin, which blocks the action of endogenous HGFA (ϩHGFsc-serum ϩleupeptin). There was some activation of Met by HGFsc in the presence of leupeptin, consistent with the small but detectable amount of activated HGF in the preparations of HGFsc (Fig. 5A). These biochemical data provide additional evidence that UB cells activate and respond to the zymogen of HGF.
Serine Protease Inhibitors Inhibit Branching Morphogenesis and Nephrogenesis by Blocking the Activation of HGF-To determine the role of HGF hydrolysis during nephrogenesis, we incubated organ cultures of kidney rudiments in protease inhibitors at concentrations that inhibited the hydrolysis of HGF by ureteric bud conditioned medium in vitro (Table I; Fig. 6, D and E). We found that the serine protease inhibitor aprotinin and the serine and cysteine protease inhibitor leupeptin blocked HGF hydrolysis in vitro. Aprotinin and leupeptin in-hibited development in rudiments isolated at E11.5 or E12.5 (Fig. 7, B, D, and F, and Table I). The effect of the serine protease inhibitors was concentration dependent. No consistent effects on renal development were noted with 1 M aprotinin or leupeptin, but development was inhibited at 10 M of either inhibitor.
To characterize the effects of serine protease inhibitors on renal development, we examined ureteric bud branching, glomerulogenesis, and nephrogenesis. Leupeptin reduced ureteric bud branching to 50% and aprotinin to 54% of control (Fig. 7, C and D; Table I). Nephrogenesis assayed by staining with the proximal tubule marker Lotus tetragonolobus lectin was qualitatively reduced by leupeptin (Fig. 7, E and F). We were unable to quantitate the effects of serine protease inhibitors on proximal tubule formation because it was impossible to reliably distinguish one long, convoluted proximal tubule from another in the control kidneys. Proximal tubule length was unambiguously decreased by leupeptin in 2/3 kidneys assayed. Serine protease inhibitors reduced glomerulogenesis (Fig. 6, C and D; Table I). In kidneys isolated at E11.5, leupeptin reduced the number of glomeruli to 53% and aprotinin to 56% of control (Table I). In kidneys isolated at E12.5, leupeptin reduced the number of glomeruli to 53% of control (Fig. 7, C and D). Thus, blocking the activation of HGF by either leupeptin or aprotinin inhibited branching morphogenesis by the ureteric bud, reduced the efficiency of nephrogenesis, and significantly decreased the number of glomeruli.
To test the specificity of the effects of serine protease inhibitors on renal development, we examined the effects of E64, a cysteine protease inhibitor that did not inhibit HGF hydrolysis by ureteric bud conditioned medium (Table I). E64 had no effect on glomerulogenesis at concentrations as high as 100 M, suggesting that the observed inhibitory effects of the serine and cysteine protease inhibitor were not nonspecific. Cells incubated in high concentrations of leupeptin (100 M, Fig. 6A) or aprotinin (not shown) were viable as assessed by trypan blue exclusion (not shown). This result shows that the serine protease inhibitors are not likely to inhibit nephrogenesis by a direct toxic effect.
We used activated, heterodimeric HGF to bypass the effects of the serine protease blockade by providing the product of HGFA activity. Activated HGF rescued nephrogenesis in leupeptin-treated kidneys (Fig. 8). Leupeptin-treated kidneys had an average of 21 Ϯ 2.4 glomeruli (n ϭ 7, S.E.), which was 53 Ϯ 6% of the number of glomeruli in untreated control kidneys (40 Ϯ 5, n ϭ 7, S.E.). Kidneys treated with leupeptin and activated HGF had an average of 34 Ϯ 5 glomeruli (n ϭ 7, S.E.), or 84 Ϯ 11% of the number of glomeruli in untreated controls. There was no significant difference between the number of glomeruli in control kidneys compared with the number of glomeruli in kidneys rescued with activated HGF (p ϭ 0.29). The number of glomeruli in leupeptin-treated kidneys was significantly decreased compared with either control (p ϭ 0.015) or HGF-rescued kidneys (p ϭ 0.04). These data suggest that loss of activated HGF is responsible for the effects of leupeptin on renal development.

DISCUSSION
In this study, we present the sequence of the murine HGFA gene. We showed that it is expressed solely by the ureteric bud in early kidney development. Using an immortalized ureteric bud cell line, we demonstrated that the ureteric bud expresses HGFA protein and have characterized the HGF hydrolyzing activity and inhibitor profile of ureteric bud derived HGFA. In the developing kidney, HGFA protein was secreted into the mesenchyme around the ampulla at the tips of ureteric bud in proximity to both its substrate, HGF, and the receptor for activated HGF, c-Met. Finally, we demonstrated, using inhibitors of HGFA, that activation of HGF by HGFA is required for renal development in vitro.
The identified cDNA is likely to be the murine homolog of HGFA because it is extremely similar to the sequence of the human gene. BLAST searches of the GenBank TM data base with the nucleic acid sequence of HGFA retrieved only the human sequence and not the sequences of other closely related molecules, for example, factor XII, urokinase, or the HGFA-like hyaluronin-binding protein. All the identified functional motifs in murine HGFA are conserved, with the majority of amino acid substitutions clustering in the amino-terminal domain between the signal sequence and the fibronectin II motif and between the kringle and catalytic domains, with the exception of the protease cleavage site, which was completely conserved. These data suggest that all of the identified motifs are important for the biologic function of HGFA. Interestingly, one of the clones we sequenced (mj50f03) had a 666-nucleotide in-frame deletion at the 5Ј end. This deletion removed both fibronectin repeats and the first epidermal growth factor repeat but preserved the signal sequence, the kringle and the catalytic do-mains. This transcript could yield a functional enzyme with altered matrix binding properties. Mj60f03 was isolated from a mouse embryo library, raising the possibility that there is developmentally regulated alternate splicing of the HGFA gene.
Immortalized ureteric bud cells expressed HGFA message and protein. The active form of HGFA is a disulfide-linked TABLE I Effect of serine and cysteine protease inhibitors on HGF hydrolysis by UB conditioned medium and on kidney development in vitro Kidney rudiments were dissected on embryonic day 11.5 (E11.5) when the ureteric bud had contacted the metanephric mesenchyme and branched once. Rudiments were grown in serum-free culture in the presence of inhibitors at the concentrations listed below for 4 days. The rudiments were then fixed and stained for glomerular markers (WT1 or peanut lectin) and ureteric bud markers (TROMA-1 or Dolichos bifloris lectin). Two or three rudiments were analyzed for each experiment, ϮS.E.  6. UB cells produce activated HGF from HGFsc and respond to activated HGF by scattering. A, UB cells were incubated for 16 h in serum-free medium alone (no HGF), 50 ng/ml activated HGF (activated HGF), 50 ng/ml HGFsc (scHGF), 50 ng/ml HGFsc ϩ 100 M leupeptin, or 50 ng/ml HGFsc ϩ 10 M leupeptin. The presence of the active heterodimeric form of HGF in the activated preparation of HGF was verified by immunoblotting. Activated HGF and HGFsc both caused scattering of UB cells. Leupeptin at 100 and 10 M blocked scattering caused by HGFsc, while 1 M leupeptin was not as effective in blocking scattering. Leupeptin had no effect on the scattering induced by activated HGF. B, the HGF receptor met was immunoprecipitated from UB cells after overnight incubation in serum-free medium. Parallel cultures were left untreated or incubated with fresh medium containing 10% fetal bovine serum and 50 ng/ml HGFsc. HGFsc with or without leupeptin were added to additional cultures to yield final concentrations of 50 ng/ml HGF and 10 M leupeptin. Met was immunoprecipated from cell lysates. Immunoblots of precipitated Met were probed successively with anti-phosphotyrosine (IB: PY) and then anti-Met (IB:Met). The molecular mass of the polypeptide was 145 kDa.  A and B), by staining for the markers TROMA-1 (ureteric bud specific cytokeratin, red) and WT1 (induced mesenchyme and intense glomerular staining, green) (C and D), and by staining for WT1 (red) and LTA (L. tetragonolobus lectin, proximal tubule, green) (E and F). A and B, leupeptin caused dysmorphic renal development and reduced the size of the kidney. C and D, leupeptin reduced branching morphogenesis by the ureteric bud and reduced the number of glomeruli. Arrowheads show representative glomeruli. E and F, the length of the proximal tubules was greatly reduced by leupeptin. We were unable to count the proximal tubules in control samples because it was difficult to identify single tubules reliably due to their convolutions. heterodimer that is generated from an inactive 99-kDa precursor by a serine protease (13,14,25). The protein secreted by ureteric bud cells was identified as a 34-kDa polypeptide on reducing gels. We used an antibody was raised against a peptide from the 34-kDa catalytic domain. One possible interpretation of this result was that the peptide antibody, which was raised against a peptide immediately adjacent to the cleavage site, could recognize only the activated form of HGFA. However, this antibody recognized both the 99-kDa zymogen and the 34-kDa catalytic domain in murine plasma. Therefore, the zymogen was not present in the ureteric bud conditioned medium. Addition of the inactive zymogen of HGF to ureteric bud cells without serum caused cell scattering, indicating that HGFA was secreted and was active. These data suggest that the ureteric bud not only secretes but also activates HGFA. Further evidence supporting the local activation of HGFA by the ureteric bud comes from the localization of HGFA in the developing, avascular kidney. The zymogen of HGFA binds weakly to heparin, while the activated form binds quite strongly (13). Thus, the observed concentration of HGFA protein around the ureteric bud suggests that HGFA was activated soon after it is secreted and therefore did not diffuse far from the ureteric bud. HGFA is activated by thrombin in vitro (25). Prothrombin message has been identified in human and rat kidneys, raising the possibility that the ureteric bud also secretes and activates prothrombin (26 -28).
We found that renal development was impeded by protease inhibitors that block HGFA activity. These effects were unlikely to be secondary to cell death produced by the inhibitors because tissue culture cells were viable in log fold higher concentrations of the inhibitors. Serine protease inhibitors blocked renal development at the same concentrations that blocked HGF hydrolysis by ureteric bud-derived HGFA. A cysteine protease inhibitor had no effect on renal development and no effect on HGF hydrolysis by HGFA, even at log fold higher concentrations. These results show that the effects of serine protease inhibitors were due to inhibition of serine protease(s) not to a nonspecific effect of protease inhibition in general and suggest that the protease involved was HGFA.
The effect of serine protease inhibitors was most likely due to the inhibition of HGF activation. First, the results of serine protease inhibition-reduced ureteric bud branching, inhibition of glomerulogenesis, and nephrogenesis are identical to the results of inhibition of HGF signaling by neutralizing antibody (7). Second, the effects of serine protease inhibition were overcome by adding the activated form of HGF. This result suggests that, at least for the parameters assayed, activation of HGF is the major role of serine proteases in renal development. Our data do not rule out a more complex situation in which serine protease inhibitors block some critical, non-HGFA activity that acts upstream of HGF and that the rescue of kidney development by activated HGF occurs irrespective of the inhibition of HGFA. However, this is not the simplest interpretation of the available data.
Several lines of evidence suggest that HGFA is the most likely candidate for the target of serine protease inhibitors in developing kidneys. First, HGFA is expressed during early renal development when our assays were performed and is a potent activator of HGF. Second, concentrations of inhibitor that blocked HGFA in vitro blocked nephrogenesis in organ culture. Third, the inhibitor profile for both HGFA blockade and inhibition of nephrogenesis was identical. However, other members of the plasminogen activator family, coagulation factor XII, tissue plasminogen activator (tPA), and urokinase (uPA), are also weak activators of HGF (11,15,16). An HGFAlike molecule, PHBP, has recently been identified and could potentially, based on its sequence, activate HGF (22,23). Two of these enzymes, tPA and uPA, are also expressed in developing kidney (29). Both plasminogen activators were detected at embryonic day 15.5 and were highly expressed, particularly uPA, by embryonic day 17.5 in the developing mouse kidney. Leupeptin does not block HGF hydrolysis by activated uPA, but it does block the activation of uPA by plasmin (11,30). Therefore, leupeptin could indirectly interfere with HGF activation by plasminogen activators by blocking their activation. Our data suggest that HGFA is the most likely target of leupeptin and aprotinin, but do not rule out a role for uPA and tPA and possibly PHBP in the activation of HGF during early renal development.
HGF and met are important in a variety of morphogenetic processes, including those of kidney, lung, mammary gland, salivary gland, liver, limb bud patterning, and migration of myogenic precursors (7,17,18,(31)(32)(33)(34). There is an absolute requirement for HGF and met in liver, limb bud, and placental morphogenesis, because development of these organs is defective in mice with targeted mutations in either gene (17,18,31). Despite the extensive in vitro evidence suggesting a role for HGF in salivary gland, lung, and kidney morphogenesis, development of these organs was reportedly normal in HGF and met null mice although the fetuses were small. Our data are consistent with the results of the gene targeting experiments in that some degree of branching morphogenesis, glomerulogenesis, and nephrogenesis did occur in the leupeptin-treated kidneys. The quantitative effects of HGFA blockade suggest that the small size of HGF and met null kidneys could have been a primary effect of the null mutation rather than a secondary effect of placental insufficiency.
Surprisingly, in view of the requirement for HGF in myogenic precursor migration, placental development, and liver morphogenesis during embryogenesis, we found that HGFA was not highly expressed in the mouse embryo and could not be detected on Northern blots. This result suggested that HGFA might be expressed at low levels and in a spatially regulated manner during development. Our data from the developing kidney suggests that this is indeed the case. We found HGFA only in the ureteric bud and in the immediately adjacent mesenchyme by protein staining and detected HGFA message in the ureteric bud by RT-PCR. A similar situation exists in the developing gastrointestinal tract, where RT-PCR was used to demonstrate that HGFA was expressed in the epithelium but not the mesenchyme (35). The identification of EST's for HGFA in embryonic mouse provides additional evidence that HGFA is expressed in the developing embryo.
In both the developing kidney and intestine, HGF was present only in the mesenchyme and HGFA was confined to the FIG. 8. Activated HGF rescues kidney development from inhibition by leupeptin. Kidney rudiments were dissected on embryonic day 12.5 and incubated in serum-free medium (control), in serum-free medium with 10 M leupeptin (leupeptin), or in serum-free medium with 10 M leupeptin and 50 ng/ml activated HGF (leupeptin ϩ HGF). Glomeruli were counted after 4 days in culture by staining with antibody to WT1. The results were normalized to the number of glomeruli in control cultures to allow comparison between litters. Seven or more rudiments from two different litters were analyzed for each condition. For this experiment, p Ͻ 0.05 for leupeptin versus HGF ϩ leupeptin, p ϭ N.S. for HGF ϩ leupeptin versus control, and p Ͻ 0.002 for leupeptin versus control. adjacent epithelium. We found that HGFA was deposited around the ureteric bud, particularly around the ampulla. The activated form of HGF binds strongly to heparin (13,36), so that activated HGF is unlikely to diffuse far from the site of activation. This observation suggests that the ureteric bud creates a local gradient of HGF activity and Met signaling. In preliminary experiments, we have identified both HGF activator inhibitor-1 and -2 (HAI-1 and HAI-2) in the ureteric bud (37)(38)(39). 2 Both HAI-1 and HAI-2 exist in membrane-anchored and soluble forms, so that there is additional potential for spatial regulation of HGFA activity in the developing kidney. These data strongly suggest that the activation of HGF is tightly controlled during renal development. As HGF-Met interactions play important developmental roles in a variety of organ systems, it is likely that similar spatio-temporal control elements are broadly expressed throughout development. Our data suggest that activation of the HGF/Met signaling pathway is regulated during embryogenesis by serine protease cascades whose activity and localization are highly regulated. The expression of HGF and its activator in adjacent structures suggest that gradients of HGF activity are formed during development. The creation and shape of such gradients could play an important role in pattern formation in HGF-responsive tissues.