scalier factorandhepatocyte growth factor: activities...

Vol. 3, 1 1-20, January 1992 Cell Growth & Differentiation 11

Scalier Factor and Hepatocyte Growth Factor:Activities, Properties, and Mechanism’

Madhu Bhargava, Ansamma Joseph, Jaromir Knesel,Ruth Halaban, Yuan Li, Susana Pang, ltzhak Goldberg,Eva Setter, Maribeth A. Donovan, Reza Zarnegar,George A. Michalopoulos, Toshikazu Nakamura,Donna Faletto, and Eliot M. Rosen2

Division of Radiation Oncology, Long Island Jewish Medical Center,

New Hyde Park, New York 1 1042 EM. B., A. J., Y. L., S. P., I. G.];

Departments of Therapeutic Radiology [J. K., E. S., M. A. D., E. M. R.];and Dermatology ER. H.], Yale University School of Medicine, NewHaven, Connecticut 06510; Department of Pathology, Duke UniversityMedical Center, Durham, North Carolina 27710 ER. Z., G. A. M.];Department of Biology, Faculty of Science 33, Kyushu University,Fukuoka 812, Japan [T. N.]; and National Cancer Institute, FrederickCancer Research Center, Frederick, Maryland 21702 ED. F.]

Abstract

Scatter fador (SF) was first identified as a fibroblast-derived protein which disperses (i.e., “scatters”)cohesive colonies of epithelium. SF-like proteins werefound in human smooth muscle cell conditionedmedium, amniotic fluid, and placental tissue. SFsmarkedly stimulate migration of epithelial, carcinoma,and vascular endothelial cell types at picomolarconcentrations. Hepatocyte growth factors (HGFs) wereoriginally described as platelet- and serum-derivedproteins which stimulate hepatocyte DNA synthesis.Partial amino acid sequence data for mouse and humanSFs indicate significant homology with HGFs. We usedbiological, biochemical, and immunological assays toevaluate and compare the adivities, properties, andmechanisms of adion of mouse SF, human SF(fibroblast or placenta derived), and recombinanthuman HGF (hrHGF). We report the following findings:(a) mouse SF exhibits species-related differences inbiological adivities relative to the human fadors; (b)human SF and hrHGF show significant overlap inbiological adivities (i.e., hrHGF stimulates motility ofmultiple normal and carcinoma cell types, whereashuman SF stimulates DNA synthesis in several normalcell types); (c) the three fadors contain commonantigenic determinants; and (d) all three proteinsstimulate rapid phosphorylation of tyrosine residues onthe c-met protooncogene protein produd (the putativereceptor for HGF) and on another protein with Mr1 10,000. A few biological and immunological differ-ences between human SFs and hrHGF were observed.These may refled minor variations in amino acid Se-

quence or posttranslational modification related to thesources of the fadors. Taken as a whole, our findingssuggest that by structural, fundional, immunological,and mechanistic criteria, human SF and human HGFare essentially identical.

Introdudion

SF3 was originally characterized as a heat-labile proteinsecreted by cultured fibroblasts which causes cohesiveepithelial colonies to spread and to “scatter” into individ-ual cells (1, 2). SF-like proteins were later identified inarterial smooth muscle cell conditioned medium (3, 4),human amniotic fluid (5), and human placental tissueextracts (5). SFs scatter carcinoma cells (4, 5) and stimu-late migration of epithelial (6-8), carcinoma (4), andvascular endothelial (7-9) cell types. Mouse (8, 10) andhuman (5, 1 1) SFs were purified and identified as heter-odimers of heavy (58 kDa) and light (31 kDa) disulfide-linked subunits. HGFs are heterodimeric proteins whichstimulate DNA synthesis in primary hepatocyte cultures(12-14). High concentrations of HGF are present in ratplatelets (12) and in human sera from patients with ful-minant liver failure (13, 15). Human and rat HGFs werecloned, and their complete amino acid sequences weredetermined from the cDNA base sequences (16-18). RatHGF was found to be about 91% identical with humanHGF, and both factors exhibited about 40% identity withhuman plasmmnogen. Partial amino acid sequence data

derived from tryptic peptides from mouse (8, 19) andhuman (11) fibroblast-derived SFs indicate significant se-quence homology with rat and human HGFs. Ninetyamino acids sequenced from seven fragments of mouseSF corresponding to different portions of the heavy (a)and light (;3) subunits of HGF showed 95.6% (86/90) and88.9% (80/90) identity with rat and human HGFs, re-spectively (Table 1). We reported that several amino acidsequences from human placental SF did not share ho-mology with HGF. However, until the complete Se-quence for placental SF is available, we cannot be certainwhether these partial sequences represent SF or acontaminant.

A HGF-life broad-spectrum mitogen was recently pu-rified from human fibroblast culture medium and iden-tified as a single-chain 87 kDa polypeptide (20). Its codingsequence contained a 5-amino acid deletion and 14single amino acid substitutions relative to the humanHGF sequence of Nakamura et a!. (16). The 145 kDa f.�

Received 7/15/91.1Supported in part by the USPHS (ROl CA50516), the American CancerSociety (BE-7), and the Finkelstein Foundation at Long Island JewishMedical Center. Dr. Rosen is an Established Investigator of the AmericanHeart Association.2To whom requests for reprints should be addressed, at Department ofTherapeutic Radiology, Yale University School of Medicine, 333 CedarStreet, New Haven, CT 06510.

3 The abbreviations used are: SF, scatter factor; kDa, kilodalton(s); HGF,hepatocyte growth factor; cDNA, complementary DNA; hr-, humanrecombinant; hf-, human fibroblast; hp-, human placental, m-, mouse;MDCK, Madin-Darby canine kidney; BBEC, bovine brain endothelialcells; BAEC, bovine aortic endothelial cells; EGF, epidermal growth factor;HPTA, hepatopoietin A; ELISA, enzyme-linked immunosorbent assay;pTyr, phosphotyrosine; DMEM, Dulbecco’s modified Eagle’s medium;

BSA, bovine serum albumin, SDS, sodium dodecyl sulfate.

12 Scatter Factor and Hepatocyte Growth Factor

Table 1 Comparison of amino acid sequences from mouse scatter factor to corresponding sequences from rat and human hepatocyte growth factors

Amino acids which differ in rat or human HGFs relative to mouse SF are underlined. Amino acids corresponding to unknown residues (Xs) are shownin parentheses. S’ = S or C. Mouse SF sequences 1 -6 were obtained by amino acid sequencing of tryptic peptides. Sequences 1 -4 were reportedpreviously by us (8), whereas sequences 5 and 6 are new. Sequence 7, which represents the NH2-terminal amino acid sequence of the fi subunit ofmouse SF, was obtained by Gherardi and Stoker (19). Sequences for rat and human HGFs were deduced from the cDNA sequences (16, 18).

. . HGF amino acidFactor Amino acid sequence number HGF subunit

1. Mouse SF V G Y E S E I P K 375-383 a (kringle)

RatHGF - . .�.9. .

HumanHGF - - � �

2.MouseSF VTLNESELCAGAEK 653-666

Rat HGFHumanHGF !

3. Mouse SF N P D G A E S P X X F 357-367 a (kringle)RatHGF (W)(d)-HumanHGF - - - - S - - -(W)(d)-

4. Mouse SF X E E G G P K X F T S ‘ N X X V 184-197 a (kringle)RatHGF (G) W(C)- - (S).(P) (E)-

HumanHGF (G) W(c). (S).(P)(E)

5.MouseSF XEGDTTPTIVNLDHPV 470-485 a

RatHGF (C)

HumanHGF (C)

6.MouseSF XXGXEFDLYENXXYI 112-126 a

RatHGF (G)(F)-(H) (K)(D).

HumanHGF (E)(F).(H) (K)(D)-

7.MouseSF V V N G I P T Q I T V 6 X M V S L- 496-519

LYRNXHI

RatHGF (W)- - - -

K- . .(K)

HumanHGF R - N - .(W). I - -

�. . .(K)- -

subunit of the c-met protooncogene protein product, atransmembrane receptor-like tyrosine kinase, was re-cently identified as a cell surface receptor for this HGF-like mitogen in B5/589 human mammary epithelial cells(21). In this study, we compared mouse SF, human SF,and human HGF to determine whether the similaritysuggested by the amino acid sequence homology isreflected in the functional and immunological propertiesof the protein.

ResultsScatter Assays. hrHGF showed a spectrum of scatteractivity similar to hfSF and hpSF (Table 2). All three factorsscattered MDCK cells, C2 and H56 rat hepatoma cells,and four of five human carcinoma lines. mSF scatteredMDCK, EMT6, C2, and H56 cells but only one humancarcinoma line. Maximal HGF scatter responses ap-peared to be similar to those of the SFs (Fig. 1). Thespecific scatter activity of hrHGF determined by MDCKserial dilution assay was about 5 units/ng, similar to thatof purified mSF (8). Thus, scatter assays distinguishedbetween mouse and human factors but did not distin-guish between human SFs and human HGF.

Migration Assays. Migration-stimulating activities ofmSF, hfSF, and hrHGF were measured in six differentcell lines using microcarrier bead assays (Fig. 2). Dose-response curves were plotted so that the abscissa rep-resents the same number of MDCK scatter units/mI foreach factor. All three factors stimulated MDCK migration,with similar dose-response curves; maximal migration

(2.2-2.4-fold over control) was observed at 100 units/mI(Fig. 2A). YaOvBix2NMA (Fig. 2B) and A253 cells (Fig.2D) were about equally sensitive to hfSF and hrHGF butwere considerably less responsive to mSF, consistentwith scatter assays. In contrast, C2 rat hepatoma cellswere more sensitive to mSF than to either human factor(Fig. 2C). C2 migration was stimulated 11-fold at 10 units/ml by mSF but only 5-fold at 50-100 units/mI by hfSF orhrHGF. Although MDCK and carcinoma cell lines re-sponded similarly to hfSF and hrHGF in scatter/migrationassays, differential responses were observed in two endo-thelial cell lines (Fig. 2, E and F). Both human factorsinduced 3-4-fold increases in migration. However, max-imal responses occurred at lower concentrations of hfSF(�1 unit/mI) than hrHGF (20 units/mI). Maximal re-sponses to mSF required 10 units/mI, but migration wassignificantly increased (2.7-3.1-fold) at 1 unit/mI. In aseparate assay, hpSF induced maximal BBEC migrationat 2 units/mI, as compared with 50 units/mI for hrHGF(data not shown). Thus, vascular endothelial cells appearto be more sensitive to low concentrations of SFs thanto hrHGF.

Chemoinvasion Assays. mSF, hpSF, and hrHGF allstimulated invasion of MDCK cells through basementmembrane (Fig. 3A). Little or no invasion was observedin the absence of factors. As with migration assays, bothmSF and hrHGF stimulated invasion of C2 cells, but mSFwas more active than the human factor (Fig. 3B). Thetumor-promoting phorbol ester 4�3-phorbol 12-myristate1 3a-acetate, which stimulates migration of MDCK cells

Cell Growth & Differentiation 13

4 M. Bhargava, A. Joseph, J. Knessel, et al., unpublished data.

Table 2 Comparison of scatter activities of mSF, hfSF, hpSF, andhrHGF on various epithelial and carcinoma cell lines

Serial dilution scatter assays were performed as described in ‘Materials

and Methods.” Concentration ranges studied were 0.05-100 MDCKscatter units/mi for each SF and 0.01-20 ng/ml for HGF. +, positivescattering (colony spreading and cell separation); -, no scattering.

Cell line (cell type) mSF hfSF hpSF hrHGF

MDCK (canine kideny epithe- + + + +

hum)

EMT6 (mouse mammary carci- + - - -

noma)C2 (rat hepatoma) + + + +

H56 (rat hepatoma) + + + +

YaOvBix2NMA (human ovary - + + +

carcinoma)A253 (human squamous carci- - + + +

noma)FaDu (human squamous carci- - + + +

noma)MCF7 (human breast carcinoma) - - - -

BT-20 (human breast carcinoma) + + + +

and a variety of carcinoma cell types (22), did not stim-ulate C2 invasion in the absence or presence of mSF.Human SFs and hrHGF stimulated invasion of YaOv-Bix2NMA cells, whereas mSF had little or no effect (Fig.3C), consistent with scatter and migration assays. Thetime course of invasion of mSF-stimulated MDCK cellswas sigmoidal in shape, with peak invasion observed at4-5 days (Fig. 3D). This finding may reflect: (a) a pro-longed effect of the factor on cell invasiveness; and (b)cell migration through “tunnels” in the Matrigel createdby previously invaded cells.

Cell Tracking Assays. MDCK cells were tracked underthe following conditions: (a) control (no addition); (b)hpSF (10 units/mI); (c) mSF (10 units/mI); and (d) hrHGF(10 units/mI). A cell was defined as “fast moving” or“slowly moving” if its average velocity was greater or lessthan the global average velocity of all 480 cells tracked.Although not all control cells were slowly moving andnot all treated cells were fast moving, there was a dra-matic increase in the random two-dimensional move-ment of isolated cells in response to each of the threefactors (Table 3). Preliminary studies suggest that thepopulation of MDCK cells consists of a mixture of SF-responsive and nonresponsive clones (data not shown).

Hepatocyte DNA Synthesis. Mouse SF stimulated DNAsynthesis in rat hepatocytes as effectively as HGF orepidermal growth factor (EGF) (Table 4). Chicken anti-serum to rabbit HPTA, a HGF-Iike serum protein (14),neutralized 60-70% of the SF- or HGF-stimulated syn-

thesis. (The same antiserum also recognized mSF in anELISA assay.) A combination of HGF and SF producedsomewhat less stimulation than either factor alone. LikemSF, hpSF also markedly stimulated rat hepatocyte DNAsynthesis (data not shown). Maximal and half-maximalDNA synthesis rates were observed at 2.5 and 0.25 ng/ml mSF, respectively. hrHGF gave corresponding valuesofabout 10-15 and 3 ng/ml,4whereas HGF purified fromplasma of patients with liver failure gave values of 8 and2.5 ng/mI (15). Neither SF (mSF, hpSF) nor hrHGF stim-ulated DNA synthesis in YaOvBix2NMA human ovarian

carcinoma cells, C2 rat hepatoma cells, or BBEC endo-thelial cells (data not shown).

SF/HGF Immunoblots. Immunological reactivities ofnonreduced mSF, hpSF, and hrHGF were comparedusing four different antibodies (Fig. 4). Anti-hrHGF anti-serum recognized all three factors. This antiserum alsoreacted with both heavy and light subunits of each factor,although staining was relatively weak (data not shown).Anti-hpSF monoclonal 23C2 also recognized all threefactors. However, a second monoclonal to hpSF (1OC1 1)recognized hpSF but not mSF or hrHGF. Antiserum tomSF oligopeptide (Ab148) recognized mSF and hpSF butdid not recognize hrHGF. Rabbit antiserum to hpSFrecognized all three factors, but the reaction to mSF wasconsiderably weaker than the reactions to hpSF andhrHGF (data not shown). The two monoclonals and

AB148 showed faint or no reaction with the individualfactor subunits. Absent or weak immunoreactivity of anti-HGF antibodies with reduced subunits relative to non-

reduced protein has been reported (23, 24). This couldbe related to conformational changes in antigenic deter-

minants due to reduction.Protein Phosphorylation Assays. We studied several

cell types to determine whether protein tyrosine phos-phorylation and, in particular, phosphorylation of thec-met protooncogene protein product were correlatedwith response to SFs and HGF. c-met protein is a heter-

odimer consisting of 50 kDa a and 145 kDa $ subunits.Only the � subunit is phosphorylated on tyrosine (21).Anti-pTyr immunoblots of reduced anti-pTyr immuno-precipitates from YaOvBix2NMA cells revealed: (a) adiffuse band at 116 kDa present in all lanes; and (b)

bands above and below at 145 kDa and 110 kDa, re-spectively, present in cells treated with mSF, hpSF, andhrHGF, but not in control cells (Fig. 5A). Under nonre-ducing conditions, factor-induced bands were observedat 190 and 110 kDa but not at 145 kDa, suggesting thatthe 145 kDa (reduced) band could represent the � sub-unit of c-met protein (data not shown). Anti-pTyr immu-noblots of anti-c-met immunoprecipitates confirmed thatthe fi subunit of c-met was phosphorylated in responseto all three factors but was not phosphorylated in re-sponse to EGF or basic fibroblast growth factor (Fig. SB).Phosphorylation of p145 (c-met) as well as p1 10 and aprotein with Mr 180,000 occurred very rapidly (within 1mm of hpSF stimulation), peaked by 5-10 mm, decreasedby 30 mm, and declined to baseline within several hours(Fig. 6). Decreased c-met phosphorylation after 30 mmmay reflect a combination of dephosphorylation andreceptor internalization and degradation. A dose-re-sponse for hpSF-induced phosphorylation in YaOv-Bix2NMA showed that c-met and p110 phosphorylationwas detectable at �100 units/mI (but not at 1 and 10units/mI) and increased up to 1000 units/mI, the highestdose studied (Fig. 7).

Similar findings were obtained with EMT6 mousemammary carcinoma cells. mSF, hpSF, and hrHGF allinduced rapid phosphorylation of c-met and severalother proteins, including p1 10 and p85. The time courseand dose-response for mSF-induced c-met phosphoryl-ation in EMT6 were similar to those for hpSF-stimulatedYaOvB1x2NMA (data not shown). Both hpSF and HGFstimulated DNA synthesis, invasion through Matrigel, and

A

B

,� -�------�

D

I S

0

iO

Fig 1. Scattering of MDCK epithelial )A-D) and C2 rat hepatoma cells )E-H) by SFs and HGF. One-day-old colonies were incubated in 96-well plateswith serial dilutions of factors for 20 h. Wells which showed maximal scattering were photographed (x 400). Concentrations of factors were 3-12 units/ml. A and E, controls; B and F, mouse SF; C and C, human fibroblast SF; 0 and H, human HGF.


�*��

I

,..

I,

HEPATOCYTE GROWTH FACTOR (nglml) (.)

SCATTER FACTOR (U/mI) (�,O)

B. C2 Rat Hepatoma (T = ID)

Chemoinvasion Assays

1 000 A. MDCK Epithelium (T = 2D) 1000

800 800

600 600

400 400CI200 #{149}�l� 20:

� 800C

600

400

200

-� -& -‘ -&

o�g� o:i: �u�co

�cg� �- 3 3 �. 00

-

to -

-+

0 -‘ -“ -‘ -� a.

g�gigi ��i0i

�34567

Time Days


Fig. 2. Microcarrier bead migra-tion assays comparing activities

of mSF (Ls), hfSF (0), and hrHGF(#{149})(see ‘Materials and Meth-ods”). Points are mean ± 1 SEM(bars) for triplicate assays. Targetcell lines (cell types) are as fol-lows: A, MDCK (canine kidneyepithelium); B, YaOvBix2NMA(human ovarian carcinoma); C,C2 (rat hepatoma); 0, A253 (hu-man squamous carcinoma); E,

BBEC (bovine brain endothe-hum); F, BAEC (bovine aortic en-dothelium). The specific scatteractivity of hrHGF was 5 MDCK

scatter units/ng. Data are plottedso that abscissas represent thesame MDCK scatter activity foreach factor.

U)

�0

0

Ca)

a)0.

U)

a)

0

z0I-.

cr0

Fig. 3. Chemoinvasion assays. Dif-

ferent cell lines were assayed forresponsiveness to mouse scatterfactor (MSF), human fibroblast SF(HFSH), human placental SF (HPSF),

recombinant human hepatocytegrowth factor (HGF), or the phorbolester 4$-phorbol 12-myristate 13a-acetate (PMA) in in vitro invasionassays (see ‘Materials and Meth-ods”). Target cell lines, assay incu-bation times, and factor concentra-tions are indicated. 0, time coursefor mSF-treated MDCK cells. Valuesplotted represent mean ± 1 SEM for3-5 replicate assays.

0 -‘ -‘ ..‘

g�g;� oIoi:

� �r1

C. YAOVBIX2NMAHuman Ovary CA (T = 2D)


S R. Halaban et al., manuscript in preparation.

Table 3 Summary of results of MDCK cell tracking assays

See text for description of assays

Parameter Control hpSF mSF hrHGF

No. of cells tracked 1 20 1 20 1 20 120No. of ‘fast moving” cells 23 82 53 64No. of ‘slowly moving” 97 38 67 56

cells

Average velocity (�m/min) 0.25 0.48 0.34 0.45

c-met phosphorylation in normal human foreskin mela-

Discussion

Mouse SF, human SFs, and human HGF share a varietyof biological activities. Like the SFs, hrHGF stimulatesscattering, migration, invasion, and random motility inmultiple normal and carcinoma cell types. Like HGF, SFsstimulate DNA synthesis in hepatocytes and melano-cytes. These proteins also share common antigenic de-terminants. Polyclonal antibodies to hrHGF and to hpSFas well as a monoclonal antibody to hpSF (23C2) recog-nized mSF, hpSF, and hrHGF on immunoblots. Similarly,a polyclonal antibody to rabbit serum HPTA, a proteinsimilar to HGF, reacted with mSF in an ELISA assay andneutralized mSF-stimulated rat hepatocyte DNA synthe-sis. All three factors stimulated tyrosine phosphorylationof the c-met protein product and of another protein witha molecular weight of 1 10,000. Mouse SF stimulatedphosphorylation in human cells and vice versa, despitethe relative lack of species cross-reactivity in bioassays(see Table 2 and Figs. 2 and 3). This observation couldreflect: (a) the existence of a second receptor (e.g., anisoform of c-met); (b) differences in the phosphorylationdose-responses at low, physiologically relevant doses offactors; or (c) another requirement for factor inductionof a biological response besides stimulation of phos-phorylation. Preliminary studies suggest that treatmentof several responsive human carcinoma cell lines witheither hpSF or human HGF results in marked enhance-ment of phosphatidylinositol-3’-kinase activity. P13K is aknown substrate for various tyrosine kinase growth factorreports, but the physiological consequences of its acti-vation are unclear. These findings suggest that SFs andHGF activate similar signal transduction pathways.

Nevertheless, some biological and immunological dif-ferences between mouse SF, human SFs, and humanHGF were found. hfSF and hpSF stimulated vascularendothelial cell migration at a 20-fold lower concentra-(ion than did hrHGF. As compared with the humanfactors, mSF usually showed greater biological activity onrodent cell lines and less activity on human lines, sug-gesting species-related differences between the mouseand human proteins. A monoclonal antibody to hpSF(lOCh) did not cross-react with mSF or hrHGF. Anti-serum to an mSF oligopeptide (Ab148) recognized mSFand hpSF but not hrHGF. 1OC1 1 and Ab148 immunoaf-finity columns prepared by conjugating antibodies toCNBr-Sepharose specifically absorbed hpSF and mSF

Table 4 Stimulation of rat hepatocyte DNA synthesis by mouse SF andhuman HGF

Values are mean ± range for duplicate assays. Antiserum is a chickenanti-rabbit HPTA immune serum which neutralizes rabbit and humanserum HPTA activity and human HGF activity (23).

. Factor added DNA synthesis (mean dpm..�x 10 ± range)

None 5.5 ± 0.2Epidermal growth factor (10 ng/ 35 ± 3

ml)Human HGF (10 ng/ml) 36 ± 2.5Mouse SF (10 ng/mI) 42 ± 4HGF (10 ng/mI) + antiserum 12 ± 1

(1:200)

SF (10 ng/ml) + antiserum (1:200) 11 ± 1EGF (10 ng/mI) + antiserum 34 ± 3

(1:200)HGF + SF (10 ng/ml each) 24 ± 2.2

scatter activity, respectively, allowing elution of activefactor with significant purification.4 Thus, these antibod-ies clearly recognize SFs.

HGF may exist in more than one form. One group ofworkers (14) prepared 10 �g purified HPTA from 4 I ofnormal human plasma, giving a concentration of >2.5ng/mI; another group (15, 24) reported a serum HGFvalue of 0.24 ± 0.12 ng/ml in 30 normal subjects, basedon a sandwich-type ELISA using an anti-human HGFmonoclonal as the first antibody. Estimates of serum HGFin patients with liver failure from bioassays of rat hepa-tocyte DNA synthesis (90-250 ng/ml) were 1 0-fold higherthan estimates from ELISAs (1 -40 ng/ml) (24). Moreover,HGF purified from sera of liver failure patients showedseveral bands (76-92 kDa) on nonreducing gels; elutedproteins from each band stimulated hepatocyte DNAsynthesis (15). Reduction yielded two different heavy (54and 65 kDa) and light (31 and 34 kDa) chains. Thus,human serum may contain two (or more) HGFs. Themonoclonal used in the ELISA may have detected onlyone of these factors. Alternatively, multiple bands mightreflect altered glycosylation of the same protein. In thiscase, the monoclonal did not recognize the major formof the protein. Although HGF, HPTA, and SF all bind toheparin, human HPTA is an acidic protein (p1 about 5.5)(14), whereas mouse SF (p1 9.5) (10) and rat platelet-derived HGF (12) are basic proteins. Rat HGF activity isacid labile, whereas HPTA is stable to 1 N acetic acid(14). Mouse SF activity is moderately sensitive to acid pH(75% reduction after 90 mm at pH 3-4) (8).

The few differences between human SF and humanHGF observed in this study raise four possibilities: (a) SFsand HGFs are closely related members of an HGF-SFmultigene family. [Minor differences in amino acid se-quence (<2%) can produce significant differences inprotein structure and function, as in the cases of sicklehemoglobin and the ABO blood group antigens (25)]; (b)they arise by alternative splicing at the mRNA level; (c)they arise from the same mRNA transcript but differ inintracellular posttranslational modification; (d) they areproduced as identical proteins, with observed differencesdue to extracellular modifications or alterations duringpurification. Other than possible microheterogeneities,based on structural, functional, immunological, andmechanistic data, human SF and human HGF appear tobe essentially identical proteins.

Anti HGF 10C11 23C2 ____ 148�_�__� -� ,_�-__� �

I

Fig. 4. Immunoblots of nonred-uced hrHGF (Lanes a), mSF(Lanes b), and hpSF (Lanes c) us-

ing various antibodies. Each lanecontained 100 ng of factor. An-tibodies includes rabbit poly-

clonal antiserum to hrHGF (Anti-HGF), two mouse monoclonals

to hpSF (bC, or 23C2), and rab-bit antiserum to mSF ohigopep-

tide I 148). Positions of the re-

duced molecular weight markers

are shown at the right. Each assaywas performed 3-4 times with

identical results.

4N

S

.�

2 ..11OK

�84K

66K

47K

__ 33K

24K


o � ;

C abc

Materials and Methods

Cell Lines and Culture

MDCK epithelial cells (CCL34), human squamous carci-noma cells {A253 (submaxillary gland, HTB41) and FaDu(hypopharynx, HTB43)], and human breast carcinomacells [MCF7 (HTB22) and BT-20 (HTB19)] were obtainedfrom the American Type Culture Collection, Rockville,MD. YaOvBix2NMA, a line of human ovarian adenocar-cinoma cells cloned after passage in nude mice (26), wasprovided by Dr. Barry Kacinski (Yale University Schoolof Medicine). C2 and H56 rat hepatoma cells were ob-tamed from Dr. Mary Weiss (Pasteur Institute, Paris).EMT6 mouse mammary carcinoma cells were providedby Dr. Sara Rockwell (Yale University School of Medi-cine). BBEC were provided by Dr. Peter Luckett (Harbor-

UCLA Medical Center, Torrance, CA). BAEC were pro-vided by Dr. Joseph Madri (Yale University School ofMedicine). Cells were subcultured at weekly intervals inDMEM plus 0.1 mM nonessential amino acids, 5.0 mg/mID-glucose, 100 units/mI penicillin, 100 zg/mI streptomy-cm, and fetal calf serum, as described previously (8).Serum concentrations (v/v) were 10% (MDCK, tumorcells) or 20% (BBEC, BAEC).

SF and HGF Preparations

mSF and hfSF were purified from conditioned mediumfrom ras-transformed NIH/2 3T3 fibroblasts (clone D4)and from diploid lung fibroblasts (CCD33Lu, AmericanType Culture Collection CRL149O), respectively (8). hpSFwas purified from extracts of term placental tissue, asdescribed previously (5), except that the final step wasimmunoaffinity chromatography using anti-hpSF IgG-Sepharose, rather than hydrophobic interaction chro-matography. hrHGF was obtained by expression of a

complete human HGF cDNA clone (16) in transformedC127 mouse fibroblasts; hrHGF was purified to >90%homogeneity from culture supernatants.

Assays of Biological Activity

Scatter Assay. Scatter activity was assessed by dilutionassay as described previously (2, 8). Briefly, cells wereinoculated into 96-well plates (5000 cells in 0.15 mlDMEM-10% serum) and incubated overnight to allowformation of small colonies. Factors were serially dilutedin 0.15 ml DMEM and added to cells. Cells were incu-bated at 37#{176}Cfor 20 h, stained, and examined for scat-tering. Activity at the limiting dilution for MDCK cells wasdefined as 0.5 scatter unit/mI.

Microcarrier Bead Migration Assay. This assay meas-ures migration of target cells from carrier beads to flatsurfaces (7). Briefly, cells were cultured for 2-3 days onCytodex 2 microcarrier beads (Pharmacia, Piscataway,NJ), washed twice in DMEM containing 5% serum, andcounted. Approximately 1 x iO� cells on beads in 0.5 mlDMEM-5% serum were seeded per well into 24-wellplates. Factors were added to wells, and cultures wereincubated for 1 8 h. Beads were removed by gentle rinsingwith phosphate-buffered saline, and cells that had mi-grated off the beads and attached to the plastic well

bottoms were stained and counted using a 1 Ox objective.Migration was expressed as cells per 10 fields (approxi-mately 0.2 cm2).

Chemoinvasion Assay. Invasion across porous filterscoated with reconstituted basement membrane (Matri-gel) was measured using blind well Boyden chambers(27). Lower and upper wells were separated by 13-mm-diameter Nucleopore filters (8-�tm pore size) coated with50 zg Matrigel (Collaborative Research). Factors dilutedin 0.2 ml DMEM containing 2.5 mg’mI bovine serum

A

205

.o

�0C-)

i-IL

Cl,0.

.C

IL�

I�-

.C

IL(1,

E 200-

‘5,:’

0

I!

80 -

p

50-- #{149}

t

�- c-met

“‘.-pllO

97 -

70 -

46 -

Fig. 6. Time course of protein tyrosine phosphorylation induced bytreatment of YaOvBix2NMA cells with human placental SF. Cells weretreated with SF (100 units/mI) at 37’C for varying times, as indicated, andthen assayed as described in ‘Materials and Methods.” Electrophoresiswas performed under reducing conditions. Arrows, positions of p145c-met) and p1 10.

�- c-met

...-pllO

�- c-met

Fig 5. Effects of scatter factors and hepatocyte growth factor on protein

tyrosine phosphorylation in YaOvBix2NMA human ovarian carcinomacells. Cells were sham-treated or treated with factors for 10 mm at 37”C

and assayed as described in “Materials and Methods.” Electrophoresiswas Perfornied under reducing conditions. A, antiphosphotyrosine im-munoblot of anti-pTyr immunoprecipitates. Factor concentrations were:control (0). human placental SF 200 units/mI). human HGF (50 ng/ml),

and mouse SF 200 units/mI). B, anti-pTyr immunoblot of anti-c-metimmunoprecipitates. Factor concentrations were: control (0), mSF (500

units/mI), hpSF (500 units/mI), hrHGF (100 ng/ml), EGF (200 ng/ml), andbasic fibroblast growth factor )bfGf) (200 ng/ml). Arrows, positions of

P145 (c-met) and p110.

18 Satter Factor and Hepatocyte Growth Factor

i:�I-,

�o�-

-C00

ri’s

t

IL

‘-‘J

0..c:

IL0I

�-

c

IL

LILU

IL

IL�

205 -

116-

albumin were placed in lower wells, and cells (5 x 10� in0.5 ml DMEM-BSA) were placed in upper wells. Cham-bers were incubated for 24-48 h (depending upon thecell line); filters were stained; and noninvading cells onthe upper surfaces were removed with a cotton swab.Nuclei of invading cells on the /ower surfaces of filterswere counted using a 40X objective.

Cell Tracking Assays. MDCK cells from a subconfluentstock culture were detached with trypsin to form a singlecell suspension and inoculated into 25-cm2 tissue culture

05.-

0C)

CCU)

�� oc,�EC�)�- C’)�-C�

flasks at 2-5 x i0� cells/flask. Cells were allowed toattach for 4 h at 37#{176}C,and flasks were filled completelywith fresh culture medium without or with factor (10units/mI). The DMIPS automated computer-controlledimaging system (28) was used to track 120 single cells/experimental condition, with each cell “revisited” every10 mm for 10 h. The position ofthe cell at each visit wasrecorded, and its speed (�tm/min) was computed as thedistance between successive recorded positions dividedby the time interval.

Hepatocyte DNA Synthesis Assays. Factors were as-sayed for stimulation of DNA synthesis in primary cul-tures of adult rat hepatocytes under serum-free condi-tions by continuous labeling with [3H]thymidine, as de-scribed previously (14).

Antibodies to SF and HGF

To prepare monoclonals to hpSF, a BALB/C3 mousewas given an i.p. injection of 30 jzg hemocyanmn-coupledhpSF in complete Freund’s adjuvant and then boosteds.c. with 10 �zg hpSF in incomplete Freund’s adjuvantevery 2 weeks for 10 weeks. Three weeks later, 5 �tgfactor were injected iv. The mouse was sacrificed 3 dayslater, and the spleen was excised. Hybridomas wereprepared by standard methods (29), cloned by limitingdilution, screened by immunoblotting, and recloned, re-suIting in two distinct monoclonals (23C2 and lOCh).To prepare Ab148, a mSF oligopeptide (Table 1, se-quence 2) was synthesized, and 200 zg hemocyanin-coupled peptide in complete Freund’s adjuvant wereinjected s.c. into a New Zealand White rabbit. Boosterinjections (100 �g peptide in incomplete adjuvant) weregiven every 2 weeks, with intervening bleeds to obtain


205

05.-

00

hpSF [U/mI]

00 0 0

0 0 0 0�- . v- � ‘-

116-

80 -

50 -

c-met

nm, 10 .zg/mI leupeptin, 1 m�s phenylmethylsulfonyl flu-oride, 100 zM NaF, 10 mt�i sodium pyrophosphate, 200�zM sodium orthovanadate, and 0.01% NaN3) (1 ml lysisbuffer/60-mm Petri dish). Lysates were vortexed, incu-bated at 0”C for 15 mm, and centrifuged for 5 mm at15,000 rpm. Supernatants were mixed with anti-phos-photyrosine monoclonal antibody PY2O (1:500; ICNBiomedicals, Inc.) and with 40 zI of 50% Protein A-Sepharose beads (Pharmacia) and incubated overnight at4#{176}Cin a shaker. The beads containing immune com-plexes were washed with wash buffer (20 m�i 4-(2-hydroxyethyl)-l -piperazineethanesulfonic acid, pH 7.5,150 mt�i NaCI, 0.1 Triton X-100, 10% glycerol, O.0h%NaN3, and 1 mM sodium orthovanadate) and eluted inboiling SDS sample buffer. Immunoprecipitates wereanalyzed on a 7.5% SDS-polyacrylamide gel, transferredto nitrocellulose, and immunoblotted using PY2O(1:1000). pTyr-containing proteins were recognized with1251-labeled Protein A. For immunoprecipitation of thec-met protein product, C28, a rabbit antiserum directedagainst a met COOH-terminal 28-amino acid peptide(34) was utilized (1:200).

Addendum

Fig. 7. Dose-response for human placental SF-induced tyrosine phos-phorylation in YaOvBix2NMA cells. Cells were treated at 37’C for 10mm with different doses of factor, as indicated. Anti-pTyr immunoblots

of anti-pTyr immunoprecipitates were prepared as described in ‘Materials

and Methods.” Electrophoresis was performed under reducing conditions.Arrows, positions of p145 (c-met) and p1 10,

antiserum. Rabbit antisera to hpSF (3) and to hrHGF wereprepared in a similar fashion. Preparation of chickenimmune serum to rabbit HPTA (14, 31) was described

elsewhere (23).

SF/HGF Immunoblotting

Purified factors (100 ng/Iane) were analyzed by SDS-

polyacrylamide gel electrophoresis on 12% acrylamidegels and electrophoretically transferred to nitrocellulose

membranes (32). Membranes were blocked with 3%BSA, washed, and incubated with antibodies for 2 h:either anti-hrHGF at 1:1000, ab148 at 1:500, or serum-free conditioned medium from hybridoma clones 23C2or hOChh. Membranes were then washed and incubatedwith a second antibody linked to alkaline phosphatase(goat anti-rabbit IgG for anti-hrHGF and Ab148; rabbitanti-mouse IgG for monoclonal antibodies; each at1:1000). Color was developed using the chromogenicsubstrate 5-bromo-4-chloro-3-indolyl phosphate (33).

Phosphotyrosine Immunoprecipitation andImmunoblotting

Cells at about 80% confluence were incubated over-night in serum-free DMEM to reduce basal phosphoryl-ation. Cells were then treated without or with purified

factors at 37#{176}C(see “Results”) and lysed in lysis buffer(50 m�’.i 4-(2-hydroxyethyl)-l -piperazineethanesulfonic

acid, pH 7.5, 150 mM NaCI, 10% glycerol, 1% Triton X-100, 1.5 mM MgCI2, 1 mivi ethyleneglycol bis (13-amino-

ethyl ether)-N,N,N’,N’-tetraacetic acid, 10 zg/ml aproti-

After this manuscript was accepted for publication, K. M.Weidner and colleagues (Proc. NatI. Acad. Sci. USA, 88:7001-7005, 1991) reported that the deduced amino acidsequences from MRC5 human fibroblast SF and humanHGF are identical.

Acknowledgments

We thank lonah Schwartz for expert assistance in preparation of this

manuscript.

References

1. Stoker, M., Gherardi, E., Perryman, M., and Gray, I. Scatter factor is a

fibroblast-derived modulator of epithelial cell mobility. Nature (Lond.),

327: 238-242, 1987.

2. Stoker, M., and Perryman, M. An epithelial scatter factor released by

embryo fibroblasts. I. Cell Sci., 77: 209-223, 1985.

3. Rosen, E. M., Goldberg, I. D., Kacinski, B. M., Buckholz, T., and Vinter,D. W. Smooth muscle releases an epithehial cell scatter factor which

binds to heparin. In Vitro Cell. Dev. Biol., 25: 163-173, 1989.

4. Rosen, E. M., Meromsky, L., Setter, E., VinIer, D. W., and Goldberg,

I. D. Smooth muscle-derived factor stimulates mobility of human tumor

cells. Invasion Metastasis, 10: 49-64, 1990.

5. Rosen, E. M., Meromsky, L., Romero, R., Setter, E., and Goldberg, I.Human placenta contains an epithelial scatter protein. Biochem. Biophys.

Res. Common., 168: 1082-1088, 1990.

6. Stoker, M. Effect of scatter factor on motility of epithehial cells and

fibroblasts. J. Cell. Physiol., 139: 565-569, 1989.

7. Rosen, E. M. Meromsky, L., Setter, E., Vinter, D. W., and Goldberg, I.D. Quantitation of cytokine-stimulated migration of endothelium and

epithehium by a new assay using microcarrier beads. Exp. Cell Res., 186:

22-31, 1990.

8. Rosen, E. M., Meromsky, L., Setter, E., Vinter, D. W., and Goldberg,I. D. Purification and migration-stimulating activities of scatter factor.

Proc. Soc. Exp. Biol. Med., 195: 34-43, 1990.

9. Rosen, E. M., Jaken, S., Carley, W., Setter, E., Bhargava, M., and

Goldberg, I. D. Regulation of motility in bovine brain endothehial cells. J.Cell. Physiol., 146: 325-335, 1991.

10. Gherardi, E., Gray, J., Stoker, M., Perryman, M., and Furlong, R.

Purification of a scatter factor, a fibroblast-derived basic protein whichmodulates epithehial interactions and movement. Proc. Natl. Acad. Sci.

USA, 86: 5844-5848, 1989.

1 1 . Weidner, K. M., Behrens, J., Vandekerckhove, I., and Birchmeier, W.Scatter factor: molecular characteristics and effect on invasiveness of

epithehial cells. I. Cell Biol., 1 1 1: 2097-2108, 1990.


12. Nakamura, T., Nawa, K., Ichihara, A., Kasire, A., and Nishino, T.

Subunit structure of hepatocyte growth factor from rat platelets. FEBS

Lett., 224: 331-338, 1987.

13. Gohda, E,, Tsubouchi, H., Nakagama, H., Hirono, S., Sakiyama, 0.,Takahashi, K., Miyazaka, H,, Hashimoto, S., and Daikuhara, Y.. Purification

and partial characterization of hepatocyte growth factor from plasma ofa patient with fulminant hepatic failure. J. Chin. Invest., 81: 414-419,1988.

14. Zarnegar, R., and Michalopoulos, G. Purification and biological char-acterization of human hepatoprotein A, a polypeptide growth factor

hepatocytes, Cancer Res., 49: 3314-3320, 1989.

15, Gohda, E., Yamasaki, T., Tsubouchi, H., Kurobe, M., Sakiyama, 0,Aoki, H., Niidani, N., Shin, S., Hayashi, K., Hashimoto, S., Daikuhara, Y.,

and Yamamoto, I. Biological and immunological properties of humanhepatocyte growth factor from plasma of patients with fulminant hepaticfailure. Biochim. Biophys. Acta., 1053: 21-26, 1990.

16. Nakamura, T., Nishizawa, T., Hagiay, M., Seki, T., Shimonishi, M.,Sugimura, A., Tashior, K., and Shimuzu, S. Molecular cloning and expres-sion of human hepatocyte growth factor. Nature (Lond.), 342: 440-443,1989,

17. Miyazawa, K., Tsubouchi, H., Naka, D., Takahashi, K., Okigaki, M.,

Arakaki, N., Nakayama, H., Hirono, S., Sakiyama, 0., Takahashi, K.,Gohda, E., Daikuhara, Y., and Kitamura, N. Molecular cloning and Se-quence analysis of cDNA for human hepatocyte growth factor. Biochem.Biophys. Res. Commun., 163: 967-973, 1989.

18. Tashiro, K,, Hagiya, M., Nishizawa, T., Seki, T., Shimonishi, M.,Shimizu, S., and Nakamura, T. Deduced primary structure of rat hepato-

cyte growth factor and expression of the mRNA in rat tissues. Proc. Nath.Acad. Sci. USA, 87: 3200-3204, 1990.

19, Gherardi, E., and Stoker, M. Hepatocytes and scatter factor (Corre-spondence). Nature (Lond.), 346: 228, 1990.

20. Rubin, J. S., Chan, A. M-L., Bottaro, D, P., Burgess, W. H., Taylor, W.G., Cech, A, C., Hirschfield, D. W., Wang, J., Miki, T., Finch, P. W., andAaronson, 5, A. A broad-spectrum human lung fibroblast-derived mitogenis a variant of hepatocyte growth factor. Proc. NatI. Acad, Sci. USA, 88:

415-419, 1990.

21. Bottaro, D. P., Rubin, J. S., Faletto, D. L., Chan, A. M-L., Kmiecik, T.

E,, Vande Woude, G. F., and Aaronson, S. A, Identification of thehepatocyte growth factor receptor as the c-met proto-oncogene product.Science (Washington DC), 251: 802-804, 1991.

22. Rosen, E. M., Meromsky, L., Goldberg, I, D., Bhargava, M., and

Setter, E. Studies on the mechanism of scatter factor: effects of agentswhich modulate intracellular signal transduction, macromolecule synthe-

sis, and cytoskeleton assembly. J. Cell Sci., 96: 639-649, 1990.

23. Zarnegar, R., Muga, S., Rahija, R., and Michalopoulos, G. Tissuedistribution of hepatopoietin-A: a heparin-binding polypeptide growth

factor for hepatocytes. Proc. NatI. Acad. Sd. USA, 87: 1252-1256, 1990.

24. Tsubouchi, H., Niitani, Y., Hirono, S., Nakagama, H., Gohda, E.,Arakaki, N., Sakiyama, 0., Takahashi, K., Kimoto, M., Kawakami, S.,

Setaguchi, M., Tachikawa, T., Shin, S., Arima, T., and Daikuhara, Y. Levelsof human hepatocyte growth factor in serum of patients with various liverdiseases determined by an enzyme-linked immunosorbent assay. Hepa-tology, in press, 1991.

25. Yamamoto, F., Clausen, H., White, T., Marken, J., and Hakomori, S.

Molecular genetic basis of the histo-blood group ABO system. Nature(Lond.), 345: 229-233, 1990.

26. Kacinski, B. M., Carter, D., Mittal, K., Yee, L. D., Scata, K. A.,Donofrio, L., Chambers, S. K., Wang, K-I., Yang-Feng, T., Rohrschnieder,

L. R., and Rothwell, V. Ovarian adenocarcinomas express fms-comple-

mentary transcripts and antigen, often with co-expression of CSF-1 . Am.J, Pathol., 137: 135-147, 1990.

27. Albini, A., lwamoto, Y., Kleinman, H. K., Martin, G. R., Aaronson, S.

A., Kozlowski, J. M., and McEwan, R. N. A rapid in vitro assay forquantitating the invasive potential of tumor cells. Cancer Res., 47: 3239-3245, 1987.

28. Thurston, G., Jaggi, B., and Palcic, B. Measurement of cell motilityand morphology with an automated microscope system. dytometry, 9:411-417, 1988.

29. Kohler, G., and Milstein, C. Continuous culture of fused cells secret-ing antibody of pre-defined specificity. Nature (Lond.), 256: 495-497,1975.

30. Bhargava, M. M., Li, Y., Joseph, A., Hofmann, R., Rosen, E. M., andGoldberg, I. D. Purification, characterization and mechanism of action ofscatter factor from human placenta. In: I. D. Goldberg and E. M. Rosen(eds.), Cell Motility Factors. Switzerland: Berkhauser-Verlag, 1991 , in

press.

31. Zarnegar, R., Muga, S., Enghild, I., and Michalopoulos, G. NH2-terminal amino acid sequence of rabbit hepatopoietin A, a heparin-

binding polypeptide growth factor for hepatocytes. Biochem. Biophys.Res. Commun., 163: 1370-1376, 1989.

32. Towbin, H., Staehehin, T., and Gordon, J. Electrophoretic transfer ofproteins from polyacrylamide gels to nitrocellulose sheets: procedure andsome applications. Proc. NatI. Acad. Sci. USA, 76: 4350-4354, 1979.

33. McGrady, J. A tetrazolium method for non-specific alkaline phospha-tase. Histochemie, 23: 180-184, 1970.

34. Gonzatti-Haces, M., Seth, A., Park, M., Copeland, T., Oroszlan, S.,Vande Woude, G. F. Characterization of the TPR-MET oncogene p65and the MET protooncogene p140 tyrosine kinases. Proc. NatI. Acad. Sci.USA, 85: 21-25, 1988.

scalier factorandhepatocyte growth factor: activities...

Documents