Petersen Protein Chemistry Laboratory, Department of Molecular and Structural Biology, University of Aarhus, Denmark In the search for new ligands for the plasminogen kringle 4 binding-p
Trang 1Tetranectin binds hepatocyte growth factor and tissue-type
plasminogen activator
Uffe B Westergaard, Mikkel H Andersen, Christian W Heegaard, Sergey N Fedosov
and Torben E Petersen
Protein Chemistry Laboratory, Department of Molecular and Structural Biology, University of Aarhus, Denmark
In the search for new ligands for the plasminogen kringle 4
binding-protein tetranectin, it has been found by ligand blot
analysis and ELISA that tetranectin specifically bound to the
plasminogen-like hepatocyte growth factor and tissue-type
plasminogen activator.The dissociation constants of these
complexes were found to be within the same order of
mag-nitude as the one for the plasminogen-tetranectin complex
The study also revealed that tetranectin did not interact
with the kindred proteins: macrophage-stimulating protein,
urokinase-type plasminogen activator and prothrombin.In
order to examine the function of tetranectin, a kinetic
analysis of the tPA-catalysed plasminogen activation was performed.The kinetic parameters of the tetranectin-stimulated enhancement of tPA were comparable to fibri-nogen fragments, which are so far the best inducer of tPA-catalysed plasminogen activation.The enhanced activation was suggested to be caused by tetranectin’s ability
to bind and accumulate tPA in an active conformation Keywords: tetranectin; plasminogen; hepatocyte growth factor; tissue-type plasminogen activator
Tetranectin (TN) is a homotrimeric C-type lectin [1].It was
originally purified due to its specific affinity towards the
kringle-4 domain of plasminogen (Plg) [2].Each of the three
20 kDa-monomers consists of an N-terminal
heparin-bind-ing domain, an a-helical domain involved in the
trimeriza-tion through a triple coiled coil structure, and a C-terminal
carbohydrate recognition domain responsible for the
bind-ing to Plg [3,4].Interaction between the carbohydrate
recognition domain and Plg is both lysine and calcium
sensitive, each of which practically abolished the binding
between TN and Plg [3]
TN was originally isolated from plasma [2] but it shows a
wide tissue distribution.Predominantly, TN was found in
the secretory cells of endocrine tissue like pituitary, thyroid,
parathyroid glands, and the liver, pancreas, and adrenal
medulla [5].A distinct accumulation of TN was observed in
the surrounding extracellular matrix of various carcinomas
[6–8] where it colocalized with Plg [9].Whether the TN
originated from plasma or it was produced by the
surround-ing tissue is still unknown.However, it has been established
that the TN concentration in plasma decreased parallel with
the growth of the tumour and this characteristic is considered
to be an indication of poor survival prognosis [10]
Although the biological function of TN is still uncertain, there has been some speculation.One suggestion is that TN forms a link between the extracellular matrix and Plg by linking Plg to sulphated polysaccharides, thus enabling a local tissue remodelling.Therefore, TN may be involved in events leading to the proteolysis of matrix proteins, as activated Plg is believed to play a key role in the degradation
of the extracellular matrix.Another possibility is that TN stimulates mineralization during osteogenesis [11] and may participate in myogenesis during embryonic development as well as muscle regeneration [12].Thus, the TN-employing mechanisms are likely to be generally applicable to tissue remodelling and not just a characteristic feature of tumour invasion
At the present stage TN has been reported to bind apolipoprotein(a) [13], fibrin [14], Plg, and some sulphated polysaccharides [15].Here evidence is presented of two novel TN-binding proteins: hepatocyte growth factor (HGF) and tissue-type plasminogen activator (tPA), whereas three related proteins: macrophage-stimulating protein (MSP), urokinase-type plasminogen activator (uPA), and pro-thrombin, appeared to be incapable of TN-binding Experimental procedures
Expression and purification of recombinant tetranectin Using the previously cloned TN cDNA [16] from a murine lung cDNA library as template in a polymerase chain reaction the coding sequence of murine TN was amplified and inserted into pPICZa-vector (Invitrogen, Netherlands) containing a signal peptide, a myc-epitope for tracing, and six histidine residues for purification at the N-terminal end
of the expressed protein.Additionally, the sequence of a coagulation factor Xacleaving site (Ile-Glu-Gly-ArgflGly) was inserted at the 5¢-end of the TN-sequence to facilitate
Correspondence to T.E.Petersen, Protein Chemistry Laboratory,
Department of Molecular and Structural Biology, University of
Aarhus, Gustav Wieds Vej10 C, DK-8000 Aarhus C, Denmark.
Fax: + 45 86 13 65 97, Tel.: + 45 89 42 50 94,
E-mail: tep@mbio.aau.dk
Abbreviations: TN, tetranectin; Plg, plasminogen; HGF, hepatocyte
growth factor; tPA, tissue-type plasminogen activator; uPA,
urokinase-type plasminogen activator; MSP,
macrophage-stimulating protein.
(Received 26 July 2002, revised 4 February 2003,
accepted 28 February 2003)
Trang 2cleavage of the N-terminal service peptides (myc, His6) after
protein purification.The construct was transfected into the
protease-deficient SMD1168 strain of Pichia pastoris as
described by the manufacturer
Positive clones were initially grown in shake flasks on
glycerol to gain a high cell density.Afterwards, the carbon
source was changed to 0.5% methanol, which induced
expression of TN.The fermentation was carried out as
outlined by the manufacturer.The cell free culture medium
was saturated with ammonium sulphate to 95% and
centrifuged.The precipitate was dissolved in 50 mM
Na2HPO4pH 7.4, 0.14 mMNaCl and dialyzed overnight
at 4C to decrease the salt concentration.From preliminary
experiments, the sample was expected to contain a mixture
of two TN-forms: myc-His6-TN and TN without the service
peptides (the latter one to be prevailing).Therefore, the
myc-His6-TN form was separated from the other by adsorption
on a Ni-matrix, and TN was purified as described below
The dialyzed sample was mixed with Ni-NTA-Sepharose,
loaded into a column, and washed with 50 mMNa2HPO4
pH 7.4, 0.5M NaCl.The run through and the wash
fractions were collected and dialyzed overnight at 4C to
remove salt.The obtained preparations were adsorbed on
an S-Sepharose column and washed until a stable baseline
was reached.TN was eluted with 50 mMNa2HPO4pH 7.4,
0.5MNaCl and concentrated by ultrafiltration
The identity of the purified protein was confirmed by
N-terminal sequencing and showed the sequence
GESPTPKAKK…, which compared to the native sequence
(ESPTPKAKK…) only including an additional N-terminal
glycine.The concentration of TN stocks was determined by
amino acid analysis
Ligand blot analysis using125I-labelled tetranectin
Three micrograms of bovine Plg, 4 lg human tPA, 2 lg
human uPA, 2 lg human HGF (294-HGN, R&D Systems
Europe, UK), 3 lg bovine prothrombin, and 2 lg human
MSP (352-MS, R&D Systems) were dissolved in the
Laemmli-buffer containing 2% SDS and subjected to
SDS/PAGE.Subsequent electroblotting was carried out
on Immobilon-P transfer membranes (Millipore, Bedford,
MA, USA).The membrane was blocked with 5% BSA,
0.05% Tween 20 and 1 mM EDTA in 50 mM Na2HPO4
pH 7.4 for 2 h The blot was then incubated overnight
at 4C with 125I-labelled TN corresponding to
100 000 cpmÆmL)1.TN was labelled with 125I according
to the chloramine T method.After washing 3· 15 min
with 0.05% Tween 20 and 1 mM EDTA in 50 mM
Na2HPO4 pH 7.4 the blot was dried and visualized by
autoradiography
Concentration-dependent binding assays
ELISA-trays (96-well) were coated with 100 lL 2 lgÆmL)1
bovine Plg, human tPA, human uPA, bovine prothrombin,
or monoclonal anti-human HGF (MAB694, R&D
Sys-tems) at 4C overnight.The wells were then blocked for 1 h
at 37C with 200 lL 0.5% gelatine and 1 mM EDTA
dissolved in 50 mMNa2HPO4pH 7.4, 0.14 mMNaCl.The
wells were washed briefly three times with 200 lL of
50 mMNaHPO pH 7.4, 0.14 mMNaCl, 0.05% Tween 20,
1 mMEDTA buffer between every incubation.There was
an additional step in the case of HGF when the antihuman HGF-containing wells were incubated with 100 lL of
1 lgÆmL)1human HGF for 2 h at 37C.All incubations were performed in blocking buffers.The compounds immobilized in the wells were then exposed to TN, which concentration varied from 0 to 50 lgÆmL)1in 100 lL.The incubation continued for 2 h at 37C.After washing, the wells were treated with 100 lL custom-made (DAKO, Roskilde, DK), specificity-checked rabbit anti-murine TN serum (1 : 1000) for 1 h at 37C.Then 100 lL of peroxidase-conjugated porcine anti-rabbit IgG (1 : 2000) was added and incubated for one hour at 37C.Finally, the TN-positive samples were visualized by a coloured reaction with o-phenylenediamine.The reaction was stopped with
100 lL 2MH2SO4, and the absorbance at 490 nm gave a relative content of bound TN
Plasminogen activation assay The ability of TN to enhance the plasminogen activation potential of tPA was investigated in a coupled reaction assay by measuring hydrolysis of the plasmin substrate
S-2251 (H-D-Val-Leu-Lys-p-nitroaniline).The proteolytic cleavage of the substrate resulting in release of the yellow p-nitroaniline was followed spectroscopically at 405 nm The plasminogen activation assay was carried out in
100 lL 0 1M Tris pH 7.4, 0.02% Tween 80, 5 lM TN, and 0.5 mM S-2251 at varying concentrations of bovine plasminogen (0.1–2.0 lM).The reaction was initiated by addition of 10 nM human tPA, and the appearing plasmin activity was followed for 3 min at 37C.Two independent experiments were made for each Plg con-centration
The activation of Plg and the subsequent hydrolysis of the plasmin substrate can be described in the simplest case by the following scheme:
tPAþ Plg !K1
tPA-Plg!k1
Plnþ S !K2
Pln-S!k2
where Pln represents plasmin, S and P represent the plasmin substrate S-2251 and the yellowish product P, respectively,
K1 and K2 are the Michaelis constants (Km) of the corresponding enzymes, k1 and k2 are their catalytic constants (kcat).The process was carried out at [tPA] << [Plg] and [Plg] constant during the time
of the reaction.The amidolytic activity of tPA (tPA + S fi tPA + P) can be ignored because of its low efficiency when compared with Pln
The collected data (t,p) was transformed and analyzed according to a previously published model [17] that gives a linear dependence of y on t2
y¼ y0þ vat2 where y¼ 2K2
k 2ln S0ep=K 2=ðS0 pÞ
and y0is the error in determination of the zero point (y0 0).The transforma-tion of P to y was carried out with the known values of
K2¼ 250 lMand k2¼ 1000 min)1[18] and the substrate concentration from this experiment s ¼ 500 lM.The slope
Trang 3va is equal to the velocity at a given [Plg].A number of
([Plg],va) pairs were obtained and fitted to the equation
tion of the kinetic parameters Kmand kcatof tPA towards
Plg, see [17] for details
Results
Purification of the recombinant tetranectin
The recombinant TN was expressed as a mixture of two
forms (myc-His6-TN and TN) due to occasional N-terminal
cleavage of the protein at the Xa-site in the yeast.The form
without service peptides (TN) prevailed and it was therefore
separated from myc-His6-TN by absorption of the latter on
Ni-NTA-sepharose.The run through fraction was subjected
to S-Sepharose, which enabled TN purification.Two intense
bands each with a relative molecular mass of approximately
20 kDa were visualized on SDS/PAGE (Fig.1) These
bands corresponded to TN on Western blot (data not
shown).Amino acid sequencing of the N-terminus revealed
that the myc-epitope and the histidine-tag were cleaved off
at the coagulation factor Xa-site upon secretion leaving the
recombinant TN of the desired length.An additional
cleavage occurred after Lys10 as well.From N-terminal
sequence analysis it was estimated that the purified protein
contained a 1 : 4 mixture of full-length and N-terminally
cleaved TN
Ligand blot analysis
In an attempt to identify new compounds capable of
TN-binding, several candidates were subjected to ligand blot
analysis.The proteins were chosen in the light of their
contents of kringle domains.Plg with five kringle domains
served as a positive control for the binding of TN.The
plasminogen-like growth factors HGF and MSP are very
similar to Plg in their domain structure and each of them
contains four kringle domains.Prothrombin has two kringle
domains and the structures of plasminogen activators tPA
and uPA include two and one kringles, respectively
The ligand blot analysis (Fig.1) revealed binding of TN
to Plg, tPA, and HGF, whereas uPA, prothrombin, and MSP showed no binding properties towards TN.Distinct bands of Plg and HGF were visible after a short time of exposure.The tPA band was somewhat weaker, but it became clearer after longer exposure
Concentration-dependent binding assays
To validate the results from the ligand blot analysis the interactions were tested by ELISA.The solid-phase binding assays revealed a concentration-dependent binding of TN to Plg, HGF, and tPA.As expected, no specific binding was observed for uPA and prothrombin (Fig.2).MSP was not included in this assay.Table 1 presents the summarized data from the ELISA, where the specific affinities are compared
to unspecific binding, and the measured dissociation constants Kdare listed for Plg, HGF, and tPA
Fig 1 SDS/PAGE/Ligand blot analysis The samples were reduced
prior to electrophoresis.Left: SDS/PAGE indicates two bands of TN
as a result of N-terminal cleavage.Right: Lanes 1–6 were loaded with
plasminogen, tPA, uPA, HGF, prothrombin, and MSP, respectively.
The blot shows TN-binding to plasminogen and HGF.However,
longer exposure revealed binding to tPA as well.
Fig 2 Concentration-dependent binding ofligands to TN The curves correspond to plasminogen (s), tPA (n), HGF (e), uPA (h), and prothrombin (,).Each point is the mean of quadruplicate determi-nations.In the case of uPA and prothrombin, no K d could be deter-mined.
Table 1 Summary ofTN affinity assays The column Bound TN corresponds to the amount of TN detected in the wells coated with a potential ligand; Unspecific binding corresponds to the amount of TN detected in the uncoated wells.The percentages are standardized according to plasminogen (100%).Calculation of the K d is based on Michaelis–Menten kinetics.Dissociation constants for the binding of
TN to plasminogen and tPA are based on three independent experi-ments of quadruplicate measureexperi-ments.For the TN-HGF binding, the
K d is based on one experiment of quadruplicate measurements.In the case of uPA and prothrombin, one and two independent experiments
of quadruplicate measurements were performed, respectively.
Bound
TN (%)
Unspecific binding (%) K d (l M ) Plg 100.0 ± 9.5 6.2 ± 3.2 0.33 ± 0.05 HGF 84.3 ± 8.7 11.8 ± 0.5 0.49 tPA 97.2 ± 14.5 8.7 ± 4.7 0.28 ± 0.09 uPA 14.2 ± 0.4 9.0 ± 1.0 –
Prothrombin 17.4 ± 5.8 8.1 ± 1.1 –
Trang 4Two different experiments complemented and supported
each other, thus confirming the affinities of the novel
TN-binding proteins
Plasminogen activation assay
It was found in the preliminary experiments that the
amidolytic activities of TN-Pln and TN-tPA complexes
towards the substrate S-2251 were equal to those of Pln and
tPA and that no difference was found in the uPA-mediated
activation of Plg with or without the inclusion of TN.This
validates the direct application of Eqn (1) for calculation
of va at different Plg-concentrations.The final data are
presented in Fig.3 as a plot of the reaction velocity vavs.the
concentration of Plg.The analysis reveals the kinetic
parameters of the activation of Plg by tPA in the presence
of TN: Km¼ 0.28 lMand kcat¼ 10 min)1.In the control
experiment the kcat could not be determined.However,
when approximating kcatto 10 min)1, Kmcan be calculated
to 2.5 lM
Discussion
The results presented here show evidence for TN-binding to
HGF and tPA by two independent methods, thereby
adding two kringle-containing proteins to the list of
TN-ligands.The affinity of the recombinant TN to Plg
(comparable to the one of the natural TN) demonstrates
that the recombinant protein has been correctly folded in
P pastorisand hence is relevant for the affinity assays.Lack
of interaction between TN and uPA indicates that TN does
not have a general affinity towards kringle domains
Moreover, TN does not bind MSP, a growth factor
structurally similar to HGF
The calculated Kd for Plg’s binding to TN
(0.33 ± 0.05 lM) is consistent with the previously
pub-lished dissociation constants of 0.5 lM and 0.2 lM
[3,13].The values of Kd for HGF (0.49 lM) and tPA
(0.28 ± 0.09 lM) are within the same order of magnitude pointing to the same type of binding for all three ligands.In other words, the carbohydrate recognition domain is very likely to be responsible for the binding of HGF and tPA, although this needs further clarification.Differences in Kd are insignificant and may be caused by the nature of the primary and secondary antibodies used for detection
It has long been known that TN stimulates the activating cleavage of Plg by tPA, though, no details about the process have been known [2].The present study verifies the activating ability of TN and shows that the presence of
TN increases the association between tPA and Plg by 10-fold.The values of Km and kcat calculated for the TN–tPA complex were similar to the parameters for the tPA activation of Plg in the presence of fibrinogen fragments (Km¼ 0.1 lMand kcat¼ 25 min)1) [17].This implies that
TN may act in the same way as fibrinogen fragments.The experimental data were, however, not conclusive concerning the exact mechanism behind the enhancement.Recently, we have reported that bovine tPA exists in equilibrium between two different conformations, where only a minor part of tPA-molecules can bind and cleave Plg.Fibrinogen frag-ments bind only the active form of tPA and thereby poises the equilibrium towards accumulation of the active confor-mation.The net result is an increased concentration of active tPA capable of cleaving Plg [17].An analogous mechanism was suggested for human tPA, though, conver-sion between inactive and active tPA occurred rapidly and was difficult to detect by conventional measurements.It seems possible that the TN-induced activation of tPA is similar to the one of fibrinogen fragments.Another explanation suggests that TN is able to bind the active form of tPA and Plg simultaneously and by bringing the two components together, it acts as a cofactor in the activation of Plg to plasmin.The ability of TN to accumulate by one or another mechanism the active tPA
in the extracellular matrix enables a higher control of the local plasmin activity
HGF is a mitogen for a variety of cells including epithelial and endothelial cells, melanocytes, keratinocytes, and hepatocytes.Additionally, the cytokine stimulates cell motility and morphogenesis.HGF has also been shown to take part in angiogenesis.HGF is a 90-kDa glycoprotein composed of a 60-kDa a chain and a 30-kDa b chain covalently linked by a disulphide bond.The a chain contains an N-terminal heparin-binding hairpin-loop and four kringle domains.The b chain is homologous to serine proteases, but lacks proteolytic activity due to mutations in the catalytic triad.HGF is secreted as a biologically inactive single-chain form and extracellular processing is required to obtain the active two-chain HGF.This is accomplished by a specific cleavage at Arg494-Val495, similar to the Arg-Val cleavage required for activation of Plg to plasmin [19] There is evidence that single-chain HGF can be activated along two or more pathways.In response to a tissue injury, epithelial cells produce HGF activator, a 34-kDa serine protease, that once it has been activated by thrombin, can process single-chain HGF [20].Another pathway involves the plasminogen activators tPA and uPA.It has been shown
in in vitro bioassays that tPA and uPA can convert single-chain HGF into biologically active HGF [21].Taking the activation of HGF by tPA together with the new findings
Fig 3 tPA-catalysed plasminogen activation in the presence ofTN The
curve represents a plot of the reaction velocity v a vs.the concentration
of plasminogen fitted to a Michaelis equation.Data obtained in the
presence of TN (s) and control data without TN (d) are based on two
or more measurements.The kinetic analysis revealed that TN has a
resemblance to fibrinogen fragments in respect to the enhancement of
tPA-catalysed plasminogen activation.
Trang 5that TN can both bind and enhance tPA activity, it is
tempting to hypothesize whether TN is involved in the
regulation of HGF as well.Testing this hypothesis would
seem evident.However, several unsuccessful attempts were
made to express recombinant single-chain HGF as the
commercially available HGF is in an already activated form
The colocalized expression of HGF and tPA in the murine
olfactory system tells in favour of this suggestion [22]
Acknowledgements
This work was supported by a grant from Novo Nordisk A/S with a
scholarship to UBW.
References
1 Holtet, T.L., Graversen, J.H., Clemmensen, I., Thøgersen, H.C &
Etzerodt, M.(1997) Tetranectin, a trimeric plasminogen-binding
C-type lectin Protein Sci 6, 1511–1515.
2 Clemmensen, I., Petersen, L.C.& Kluft, C.(1986) Purification and
characterization of a novel, oligomeric, plasminogen kringle 4
binding protein from human plasma: tetranectin Eur J Biochem.
156, 327–333.
3 Graversen, J.H., Lorentsen, R.H., Jacobsen, C., Moestrup, S.K.,
Sigurskjold, B.W., Thøgersen, H.C & Etzerodt, M (1998) The
plasminogen binding site of the C-type lectin tetranectin is located
in the carbohydrate recognition domain, and binding is sensitive to
both calcium and lysine J Biol Chem 273, 29241–29246.
4 Nielsen, B B , Kastrup, J S , Rasmussen, H , Holtet, T L ,
Graversen, J.H., Etzerodt, M., Thøgersen, H.C & Larsen, I.K.
(1997) Crystal structure of tetranectin, a trimeric
plasminogen-binding protein with an alpha-helical coiled coil FEBS Lett 412,
388–396.
5 Christensen, L., Johansen, N., Jensen, B.A & Clemmensen, I.
(1987) Immunohistochemical localization of a novel, human
plasma protein, tetranectin, in human endocrine tissues.
Histochemistry 87, 195–199.
6 Christensen, L.& Clemmensen, I.(1991) Differences in tetranectin
immunoreactivity between benign and malignant breast tissue.
Histochemistry 95, 427–433.
7.Høgdall, C K , Christensen, L.& Clemmensen, I.(1993) The
prognostic value of tetranectin immunoreactivity and plasma
tet-ranectin in patients with ovarian cancer Cancer 72, 2415–2422.
8 Wewer, U.M.& Albrechtsen, R.(1992) Tetranectin, a
plasmino-gen kringle 4-binding protein.Cloning and plasmino-gene expression
pattern in human colon cancer Laboratory Invest 67, 253–262.
9 De Vries, T.J., De Wit, P.E., Clemmensen, I., Verspaget, H.W.,
Weidle, U H , Bro¨cker, E.B., Ruiter, D.J & Van Muijen, G.N.
(1996) Tetranectin and plasmin/plasminogen are similarly distri-buted at the invasive front of cutaneous melanoma lesions.
J Pathol 179, 260–265.
10 Høgdall, C.K., So¨le´tormos, G , Nielsen, D , Nørgaard-Pedersen, B., Dombernowsky, P.& Clemmensen, I.(1993) Prognostic value
of serum tetranectin in patients with metastatic breast cancer Acta Oncol 32, 631–636.
11 Wewer, U M , Ibaraki, K , Schjørring, P , Durkin, M E , Young, M F.& Albrechtsen, R.(1994) A potential role for tetranectin in mineralization during osteogenesis J Cell Biol 127, 1767–1775.
12 Wewer, U M , Iba, K , Durkin, M E , Nielsen, F C , Loechel, F , Gilpin, B.J., Kuang, W., Engvall, E & Albrechtsen, R (1998) Tetranectin is a novel marker for myogenesis during embryonic development, muscle regeneration, and muscle cell differentiation
in vitro Dev Biol 200, 247–259.
13 Kluft, C , Jie, A F , Los, P , de Wit, E & Havekes, L (1989) Functional analogy between lipoprotein (a) and plasminogen in the binding to the kringle 4 binding protein, tetranectin Biochem Biophys Res Commun 161, 427–433.
14 Kluft, C., Los, P.& Clemmensen, I.(1989) Calcium-dependent binding of tetranectin to fibrin Thromb Res 55, 233–238 15.Clemmensen, I.(1989) Interaction of tetranectin with sulphated polysaccharides and trypan blue Scand J Clin Laboratory Invest.
49, 719–725.
16 Sørensen, C.B., Berglund, L & Petersen, T.E (1995) Cloning of a cDNA encoding murine tetranectin Gene 152, 243–245.
17 Johnsen, L.B., Ravn, P., Berglund, L., Petersen, T.E., Rasmussen, L.K., Heegaard, C.W., Rasmussen, J.T., Benfeldt, C & Fedosov, S.N (1998) A refined kinetic analysis of plasminogen activation by recombinant bovine tissue-type plasminogen activator indicates two interconvertible activator forms Biochemistry 37, 12631– 12639.
18 Wohl, R.C., Sinio, L., Summaria, L & Robbins, K.C (1983) Comparative activation kinetics of mammalian plasminogens Biochim Biophys Acta 745, 20–31.
19 Rosen, E.M & Goldberg, I.D (1997) Regulation of angiogenesis
by scatter factor EXS 79, 193–208.
20 Shimomura, T., Kondo, J., Ochiai, M., Naka, D., Miyazawa, K., Morimoto, Y.& Kitamura, N.(1993) Activation of the zymogen
of hepatocyte growth factor activator by thrombin J Biol Chem.
268, 22927–22932.
21 Mars, W.M., Zarnegar, R & Michalopoulos, G.K (1993) Acti-vation of hepatocyte growth factor by the plasminogen activators uPA and tPA Am J Pathol 143, 949–958.
22 Thewke, D.P & Seeds, N.W (1996) Expression of hepatocyte growth factor/scatter factor, its receptor, c-met, and tissue-type plasminogen activator during development of the murine olfac-tory system J Neurosci 16, 6933–6944.