Activation of Hepatocyte Growth Factor-Met Autocrine Loop Enhances Tumorigenicity in a Human Lung Adenocarcinoma Cell Line

Abstract

Hepatocyte growth factor (HGF) is a multifunctional cytokine with effects on the proliferation, motility, and differentiation of cells that express its receptor Met. The co-expression of HGF and Met is common among non-small-cell lung cancers, especially adenocarcinoma. However, the biologic consequences of this putative HGF-Met autocrine signaling remain speculative. We have used retroviral gene transduction technique to express high levels of HGF in the NCI-H358 lung adenocarcinoma cells that have functionally active cell surface Met receptor. The activation of autocrine HGF-Met signaling was confirmed by the induction of spontaneous cell scattering activity. Compared to the parent and control cells transduced with the retroviral vector alone, HGF overexpressing H358 cells show enhanced capacity to colonize soft agar medium and to form xenograft tumors when implanted in the subcutaneous tissue of immune-deficient mice. These effects were not accompanied by changes in their growth rate in monolayer culture condition, or in the expression of vascular endothelial growth factor. The tumors formed by HGF overexpressing cells also showed more prominent glandular cell arrangement and functional activity. This report provides the direct in vivo evidence that autocrine HGF-Met signaling plays significant roles in the growth and differentiation of human lung adenocarcinoma cells.

Introduction

Hepatocyte growth factor (HGF), also known as scatter factor (SF), is a heterodimeric protein composed of a 60-kDa α-chain and a 32- to 36-kDa β-chain [1–3]. The mature protein is formed by proteolytic processing from a 92-kDa single-chain precursor polypeptide [4]. The receptor for HGF is Met receptor tyrosine kinase that was originally identified as the protein product of c-met proto-oncogene [5]. The interaction of HGF with Met receptor exerts pleiotropic effects in vitro on cell proliferation, motility, and morphogenesis [6–8]. In vivo studies have also demonstrated that HGF plays important roles in angiogenesis [9], organogenesis [10], and neoplastic development. HGF-expressing transgenic mice have been reported to form tumors in various organs [11].

HGF is primarily produced in vitro by cultured mesenchymal cells, but most epithelial cells express the Met receptor [12]. HGF has therefore been considered mainly as a paracrine mediator of mesenchymal-epithelial cell communication [8]. The co-expression of HGF and Met, however, has been reported in many types of cancers (reviewed in Ref. [13]). Experimental generation of HGF-Met autocrine loop in rodent and human sarcoma cell lines has resulted in enhancement of tumorigenicity and metastasis in vivo [14–17], indicating that autocrine signaling of HGF-Met may contribute to tumor progression. On the other hand, enhanced tumorigenicity by HGF expression in a human breast carcinoma cell line was attributed to its paracrine function on angiogenesis [18].

Immunoreactive HGF protein can be detected in approximately two thirds of lung cancer tissues regardless of histologic type [19–22]. The Met receptor is also expressed in a significant proportion of primary tumors and cell lines of non-small-cell lung carcinoma (NSCLC) [19,21,23,24]. Immunohistochemistry has detected HGF and Met co-expression in 50% of lung adenocarcinoma [19], and low levels of HGF mRNA expression were found in approximately 50% of NSCLC cell lines studied [25]. Seven of these lines also co-expressed the Met receptor. The co-expression of HGF and Met in lung adenocarcinoma and high levels of HGF content in primary NSCLC tissues have been reported to represent poor prognostic factors [21,22], suggesting that autocrine HGF-Met signaling plays a very important role in the biologic behavior of lung cancer cells, especially adenocarcinoma. We report here the experimental evidence that supports the critical role of autocrine HGF-Met signaling in lung cancer progression.

Materials and Methods

Cell Lines

NCI-H358 is a lung adenocarcinoma cell line previously established and characterized at the NCI-Navy Hospital, Bethesda, MD [26]. Cells were routinely cultured at 37°C in 5% CO₂ atmosphere in RPMI-1640 medium (Life Technologies, Gaithersburg, MD) supplemented with 10% fetal bovine serum (R10 medium).

Generation of HGF/SF Overexpressing Cells

The pLXSN retroviral plasmid and pLXSN-HGF plasmids containing a 2.6-kb full-length HGF cDNA were respectively co-transfected with SV-Ψ-A-MLV helper virus plasmid into Cos cells [27,28]. The culture supernatants that contained these defective retroviral particles were harvested 72 hours later, filtered through 0.45-µm cellulose acetate membrane, aliquoted and stored at -80°C until use. NCI-H358 (2.5x10⁵) cells were seeded on to 60-mm tissue culture dishes (Becton Dickinson, Lincoln Park, NJ) and grown for 24 hours before retroviral infection. After cells were washed twice in Hanks' balanced salt solution (HBSS, Life Technologies), they were cultured in RPMI-1640 medium containing 10% heat-inactivated FBS, and with the viral supernatants added in the presence of 4 µg/ml of polybrene (Sigma Chemical Co., St. Louis, MO). Twenty hours later, the media were removed and the cells were washed twice with HBSS and selected with 400 µg/ml of G418 in R-10 media. Neomycin-resistant colonies were usually obtained 12 to 18 days later, and cloned using stainless-steel cloning rings.

Reverse Transcription-Polymerase Chain Reaction (RT-PCR)

Total cellular RNA was isolated from pre-confluent cells using the standard acid guanidium isothiocyanate-phenol/chloroform extraction method as previously described [24]. Three micrograms of total RNA that has been pretreated with RNase-free DNase (1 unit/µg RNA, Boehringer Mannheim Canada, Dorval, Quebec, Canada) was reverse transcribed using 0.67 mM pd(N)₆ random primer and Ready-To-Go (Pharmacia Biotech, Piscataway, NJ). Sixteen of 33-µl RT products were subjected to PCR in the presence of 1x Taq buffer, 0.2 mM dNTP, 1.5 U Taq Polymerase, and 0.15 µM each of the 5′ and 3′ primers. The forward HGF primer was 5′-CAGCGTTGGGATTCTCAGTAT-3′ (979–1000) and the reverse primer was 5′-TCCAACACGAAGAAACATAGG-3′ (1497–1518), and the expected amplified PCR product is 539 base pairs (bp) [29]. PCR was performed for 30 cycles, consisting of 1-minute denaturation at 94°C, 1-minute annealing at 53.5°C and 1.5-minute extension at 72°C. The amplification products were visualized by ethidium bromide following electrophoretic separation in 1.5% agarose gel.

Northern Blot Analysis

Twenty micrograms of total RNA samples was electrophoretically separated in 1% formaldehyde-agarose gel, then transferred onto Hybond-N membrane (Amersham, Oakville, Ontario, Canada) in 10x standard saline citrate (SSC) buffer (1xSSC contains 0.1 M NaCl and 15 mM sodium citrate). After air drying, the membrane was exposed to UV light for cross-linking. The membrane was prehybridized at 68°C for 30 minutes in QuikHyb hybridization solution (Stratagene, La Jolla, CA), then hybridized to ³²P-labeled cDNA probes using the same solution. Met mRNA was probed with the 841-bp C-terminal met cDNA fragment [30], with radioactive labeling performed using the random primer oligolabelling kit (Pharmacia Biotech). Hybridization occurred at 68°C for 18 hours in the QuikHyb solution. The membranes were then washed twice for 30 minutes each time at 60°C in 0.1xSSC containing 0.1% sodium dodecyl sulfate (SDS), then exposed to Kodak XAR-5 film at -80°C overnight. After stripping in boiling washing buffer solution, the same membrane was subsequently rehybridized with a ³²P-labelled 890-bp human VEGF cDNA probe [32] and 260-bp β-actin cDNA probes. The intensity of hybridization signals was assessed by a densitometer, and their relative values were normalized to the β-actin signals that estimated the relative amount of RNA loads.

Enzyme-Linked Immunosorbent Assay (ELISA)

Following isolation by cloning rings, each G418-resistant clone was transferred into 12-well tissue culture plates and grown to confluence. The cells were washed with HBSS and incubated in 1 ml of serum-free RPMI-1640 at 37°C for 48 hours. The conditioned media were obtained and cleared with centrifugation at 10,000g, and HGF was assayed using a double sandwich ELISA [32]. Briefly, Nunc 96-well Maxisorp plates were coated with 5 µg/ml monoclonal HGF antibody (Genentech, South San Francisco, CA) dissolved in 50 mM carbonate/bicarbonate buffer (pH 9.6). This was done by sequential incubation at 37°C for 1 hour, than overnight at 4°C. Following three-time washes with PBS buffer containing 0.1% Tween-20, (pH 7.4), each well was incubated with blocking buffer (washing buffer with 0.5% bovine serum albumin) at room temperature for 2 hours to prevent nonspecific interaction, then washed again. Ten microliters of the supernatants was diluted in HBSS to a final volume of 100 µl, then incubated in each well of Maxisorp at room temperature for 2 hours. Recombinant human HGF (Genentech) at concentrations ranging from 0 to 10 ng/ml was used as standards. After three-time washes, the wells were incubated at room temperature for 2 hours with biotinylated guinea pig HGF antibody (Genentech), washed three times, then incubated for 1 hour with horseradish peroxidase-streptavidin (Amersham Canada, Oakville, ON, Canada) conjugate at 1:2500 dilution in Ca²⁺/Mg²⁺-free PBS at room temperature. Following three further washes, 100 µl of tetramethylbenzidine dihydrochloride solution (Sigma) was added into each well, and incubated at room temperature until color changes appeared. After addition of the same volume of 0.5 M H₂SO₄, optical density was measured at 450 nm using the Titertek Multiskan MCC/340.

Western Blot Analysis

HGF production was also assayed by Western blot technique. Ten milliliters of conditioned medium was collected from the pre-confluent cultures of each cell clone grown in 100-mm tissue culture dish. The media were lyophilized and redissolved in 100 µl dH₂O. Thirty micrograms of total protein from each sample was separated in 8% polyacrylamide gel, then transferred on to a nitrocellulose membrane. Following blocking with 5% skimmed milk in Tris-buffered saline-Tween 20 (TBS-T) buffer containing 10 mM Tris-HCl (pH 7.0), 0.1% Tween-20, 2.5 mM EDTA, and 50 mM NaCl, the membrane was incubated with 1000-fold diluted goat polyclonal HGF antibody (Santa Cruz Biotech., Santa Cruz, CA) in TBS-T buffer. HGF protein was subsequently revealed by the BM Chemiluminescence Kit (Boehringer Mannheim Canada). HGF expression in tumor tissue was also assayed similarly following extraction of the total protein using a tissue homogenizer.

Met protein levels were also estimated by the Western blot technique. Pre-confluent cells in 60-mm tissue culture plates were incubated in serum-free RPMI-1640 medium for 48 hours, then washed with cold HBSS and lysed in a lysis buffer. The buffer contained 50 mM Hepes, pH 8.0, 10% glycerol, 1% Triton X-100, 150 mM NaCl, 1 mM EDTA, 1.5 mM MgCl₂, 100 mM NaF, 10 mM Na₄P₂O₇, 1 mM Na₃VO₄, and supplemented with 5 µg/ml leupeptin, 5 µg/ml aprotinin and 100 µg/ml PMSF. Three hundred micrograms of lysate protein from each clone was incubated with gentle rocking in 0.1% lysis buffer containing the mouse monoclonal Met antibody (C-28, Santa Cruz Biotech, Santa Cruz, CA) at 4°C for 1.5 hours. After centrifugation at 10,000g for 5 minutes, the precipitated immunocomplex was washed three times in 0.1% lysis buffer containing 1 mM Na₃VO₄, boiled in loading buffer plus 0.2 M dithiothreitol at 95°C for 5 minutes. The proteins were then separated in 8% polyacrylamide gel. Following transfer on to nitrocellulose membrane, the membrane was immunoblotted with C-28 Met antibody (Santa Cruz Biotech.). The relative intensity of the chemiluminescent signals was estimated using a densitometer.

In Vitro Proliferation

Ten thousand cells were seeded into each well of six-well tissue culture dishes (Life Technologies) in R-10 medium. The numbers of cells in each of triplicate wells were counted every 2 days. The population doubling time was estimated from the exponential phase of growth by a previously reported equation [33].

Anchorage-Independent Growth

This was performed as previously reported [34]. Ten thousand cells were suspended in R-10 medium containing 0.3% agar medium stacked on a 0.5% basal layer agar medium in 60-mm tissue culture plate. One milliliter of R-10 medium was added 1 day later and weekly thereafter. Each assay was performed with triplicate plates, and the cells were cultured for 50 days. Viable colonies were stained by incubation at 37°C for 24 hours in an aqueous solution containing 0.5 mg/ml β-NADH (Boehringer Mannheim Canada) and 0.5 mg/ml p-nitroblue tetrazolium chloride (Sigma). The number of cell colonies larger than approximately 100-µm diameter was counted with the aid of a dissecting microscope.

Spontaneous Cell Motility

Cells of each clone were grown to confluence in 12-well tissue culture plates. Linear scratches were made using the sterile plastic micropipette tips. Twenty one hours later, cells were fixed in a solution of methanol, acetic acid and dH₂O in 75:5:20 volume ratio. The plates were stained with 6% Giemsa solution, and cells that have migrated from wound edge were counted within 1.0-mm length. The measurements were obtained from triplicate wells.

Culture in Collagen Matrix

The collagen gel culture medium was prepared using rat tail type I collagen (Becton Dickinson, Bedford, MA) as previously described [25]. One thousand cells embedded in collagen matrix were placed in each well of 12-well tissue culture plate, and were incubated in R-10 medium for two weeks. The morphologic appearances of the colonies formed were assessed using an inverted phase-contrast microscope.

Tumorigenicity Assay

Cells grown to 80–90% confluence were dissociated with trypsin in Ca²⁺/Mg²⁺-free HBSS, washed twice with HBSS, and resuspended in serum-free RPMI-1640 medium supplemented by 10% (v/v) Matrigel (Becton Dickinson) at 2x10⁷ cells/ml concentration. One hundred microliters of cell suspension was subcutaneously injected into the neck fat pad of 5- to 6-week-old male SCID mice (Ontario Cancer Institute, Toronto, ON, Canada). The tumorigenicity of each cell line or clone except for the parental line was assayed using four to six mice. The animals were checked weekly for tumour formation. The minimum diameter of tumour that can be palpated and measured by caliper was 4 mm, and the longest and shortest diameters of tumor mass were measured. When average tumor diameter reached 10 mm, the mouse was sacrificed. The non-tumour-bearing mice were kept for 27 weeks after cell implantation before sacrifice.

Following CO₂ euthanasia, the tumor mass was removed and its dimension was measured accurately. The tumor mass was then divided into two fragments. One half was fixed in 10% buffered formalin, and the other half was snap-frozen in liquid N₂ and stored at -80°C. To check for metastasis, the lungs of these mice were also harvested. The formalin-fixed tumor and lung tissues were routinely processed for paraffin embedding, and sections were stained by hematoxylin-eosin for histologic evaluation.

Results

HGF Expression

Infection of H358 cells with pLXSN and pLXSN-HGF retrovirus supernatants yielded many G-418 resistant colonies. Twelve colonies infected with control pLXSN (PL) and 36 clones infected with pLXSN-HGF (PSF) were randomly cloned. Following ELISA screening of HGF secretion in conditioned media, two representative negative control lines (PL2 and PL4) and six HGF expressing lines (PSF2, PSF3, PSF8, PSF11, PSF17, and PSF24) were arbitrarily chosen for further studies. Although HGF was detected in conditioned media of neither parent H358 line nor any PL clones, the levels of HGF secreted by all PSF clones were higher than 4.8 ng per 10 µl of 48-hour conditioned media (Table 1). RT-PCR and Western blot assays demonstrated the mRNA expression and synthesis/secretion of HGF (Figure 1, A and B). The conditioned media of these PSF clones also induced the scattering of MDCK cells, confirming that the HGF secreted by these clones was biologically active (data not shown).

Table 1.

Phenotypic Properties of NCI-H358 Cells and their Clones Isolated after Infection by Retrovirus Generated by either pLXSN or pLXSN-HGF Plasmids.

Cell Line/Clone	Protein Expression		Doubling Time (hours)*	Spontaneous Motility (Relative to H358 Cells)†	Branching Morphogenesis in Collagen Gel‡
	HGF/SF (ng/10 µl)§	Met (Relative to H358 Cells)¶
Parent	0	1.00	17.7	1.0	-
PL2	0	1.06	20.8	5.0±1.0	-
PL4	0	1.24	22.9	2.7±1.5	-
PSF2	>5.86	0.41	43.0	49.0±8.4	++
PSF3	>5.86	0.53	22.4	16.3±2.0	++
PSF8	5.08	0.50	23.9	16.3±2.7	+
PSF11	4.84	0.57	20.5	53.7±3.0	+++
PSF17	5.37	0.79	18.4	19.3±2.6	++
PSF24	5.55	0.92	20.4	61.0±11.4	+

^*Doubling time=0.693(t-t₀)/In(N_t/N₀). (t₀=time at which exponential growth phase began, t=time at certain point, N_t=cell number at time t, and N₀=initial cell number.

^†Mean±standard error.

^‡The degree of branching extensions formed by cell colonies are estimated semi-quantitatively from - (none) to + + + (extensive).

^§HGF concentrations in the conditioned media as determined by ELISA.

^¶Relative levels as estimated by Western blot signals.

Figure 1.

Overexpression of HGF in NCI-H358 adenocarcinoma cells. (A) RT-PCR demonstrated high levels of HGF mRNA expression in clones of H358 cells transduced by the pLXSN-HGF retrovirus, but not in the parent or control PL cells transduced with pLXSN virus. The MGH-7 lung squamous cell carcinoma cell line that naturally expresses low levels of HGF was used as a positive control. (B) Western blot analyses of concentrated conditioned media showed the secretion of full length HGF by PSF but not by the parent H358 or PL control cells. The N-17 antibody detected both the 92-kDa unprocessed single chain and 60-kDa processed α-chain of HGF. (C). Northern blot analyses showing c-met and VEGF mRNA expression by parent H358, the control PL and HGF overexpressing PSF clones. Rehybridization with the β-actin cDNA demonstrated a relatively equal loading of RNA. (D) Met protein levels as detected by Western immunoblot with a Met antibody. The HGF/SF overexpressing cells tended to show reduced levels of Met receptor protein that is represented by the 145-kDa β-chain polypeptide.

Met Receptor Expression

NCI-H358 parental cell line expressed moderate level of Met mRNA and protein comparable to that of cultured normal bronchial epithelial cells [25]. The met mRNA levels were not significantly different between the PL and PSF clones compared to the parental line (Figure 1C). The Met protein levels in both PL clones were similar to that in parent line, but were reduced 21–59% in five of six PSF clones (Table 1 and Figure 1D). The VEGF mRNA expression levels in the PL and PSF clones were not consistently or significantly altered compared to the parent H358 cells (Figure 1C).

Activation of Autocrine HGF-Met Signaling Activity

We previously reported that exogenous HGF induced H358 cells to scatter in monolayer culture, and to form branching extensions in when grown in collagen matrix [25]. On plastic surface, H358 and PL cells did not show significant spontaneous scatter activity (Figure 2, A and C). In contrast, the PSF cells demonstrated 16- to 61-fold increases in their spontaneous motility compared to the parent H358 cells (Figure 2, B and D; Table 1).

Figure 2.

The activation of autocrine HGF-Met signaling in HGF overexpressing H358-PSF cells. The phase contrast appearances of PL2 (A) and PSF24 (B) cells showed the presence of spontaneous scattering activity in the PSF cells only. Scattering activity was also quantified using the wound migration assay. The PL2 cells (C) demonstrated very sparse spontaneous movement into the “wound” area. In contrast, the PSF24 cells (D) showed a brisk cellular migration.

When H358 cells were grown in collagen gel, they formed solid round colonies. An exposure to HGF resulted in the formation of irregular branching extensions [25]. The cells of PL clones also formed simple round aggregates in collagen gel, but all PSF clones demonstrated formation of branching structures (Table 1). Histologic examination of these branching cell cords did not demonstrate lumen formation, as previously described [25]. These motogenic and morphogenic results confirmed the activation of HGF-Met autocrine signaling loop in the PSF clones but not in PL clones.

In Vitro Growth Properties

Most clones of both PL and PSF except the PSF2 lines did not show significant change in their population doubling time (18.4 to 23.9 hours) compared to the parental line (17.7 hours). This indicates that overexpression of HGF in NCI-H358 did not affect cell proliferation in vitro (Figure 3 and Table 1), consistent with our previous observation that exogenous recombinant HGF did not notably change the growth rate of H358 cells in monolayer culture condition [25].

Figure 3.

The effect of HGF overexpression on the proliferation of NCI-H358 cells in monolayer culture. (A) Control parent H358 and pL clones. (B) HGF overexpressing PSF clones. Data are represented by the means±S.E.M. of triplicate experiments.

In contrast to anchorage-dependent growth on plastic surface, the anchorage-independent growth in soft agar was significantly affected by the overexpression of HGF. The parent H358 line and mixed population or clones of PL cells demonstrated very low soft agar colony-forming efficiency. Five- to tenfold higher colony-forming efficiency was demonstrated in the mixed population and four of six clones of PSF cells (Table 2).

Table 2.

Effect of HGF Overexpression on Soft Anchorage-Independent Growth and Tumorigenicity of NCI-H358 Cells.

Cell Line	Colony-Forming Efficiency in Soft Agar (%)*	Animal with Tumor/Animal Injected (%)	Mean Day±SD to Develop 10-mm Tumor (Range)	VEGF mRNA (Fold of Parent)
Parent	0.13±0.02	2/15 (13%)	137±25 (112, 162)	1
PL-pool	0.24±0.05	0/5 (0%)	NA	N.D.
PL2	0	0/6 (0%)	NA	2.225
PL4	0	0/6 (0%)	NA	1.81
PSF-pool	1.24±0.21	4/4 (100%)	73±9 (56–99)	N.D.
PSF2	1.72±0.61	4/4 (100%)	90±10 (80–100)	1.19
PSF3	0.35±0.05	6/6 (100%)	79±9 (67–88)	1.34
PSF8	1.09±0.33	4/4 (100%)	93±12 (73–100)	1.19
PSF11	1.45±0.23	4/4 (100%)	109±15 (98–119)	1.70
PSF17	0.03±0.02	4/4 (100%)	88±9 (76–98)	1.96
PSF24	1.89±0.54	4/4 (100%)	96±18 (70–112)	1.56

^*Number of colonies/number of cells seededx100.

N.D.: not done; N.A.: not applicable.

Tumorigenicity in Immune-Deficient Mice

Only 2 of 15 mice injected with the parental NCI-H358 cells formed a tumor in immune-deficient mice (Table 2). These tumors were initially detected at the 11th and 16th weeks after cell implantation, and reached approximately 10-mm diameter at the 16th and 23rd week, respectively. The pooled population of chemically selected PL cells and two isolated clones were also assayed. They all failed to develop tumor for up to 27th week after cell implantation. In contrast, the pooled population of PSF cells and all four clones tested formed tumors in all animals injected with the cells. These PSF tumors were detected at the earliest in the third to twelfth weeks, and on average reached 10-mm diameter in the eleventh to fifteenth weeks after implantation. Thus, although the control line/clones showed an overall 6% (2/26) tumorigenicity, PSF cells demonstrated 100% (30/30) tumorigenicity. Western blotting confirmed HGF expression in xenograft tumors formed by the PSF clones (Figure 4A). None of the animals with tumor growth showed either gross or microscopic metastatic deposits in their lungs.

Figure 4.

The xenograft tumors formed by the HGF overexpressing H358 cells. (A) High levels of HGF protein was found in extracts of the tumors formed by the PSF cell lines but not in the tumor formed by the parent H358 cells. (B) The H358 tumor was a poorly differentiated adenocarcinoma with focal formation of glands with small lumens. (C) Focal area of tumour formed by HGF overexpressing H358 cells showing papillary and prominent alveolar gland structures. (D) HGF overexpressing tumors also consistently formed cysts and showed many cells with cytoplasmic vacuoles containing PAS positive and diastase resistant material (arrow). (B and C: H and E, x250; D: PASD, x400).

VEGF Expression in HGF Overexpressing Cells

NCI-H358 cell line expressed relatively moderate levels of vascular endothelial growth factor (VEGF) mRNA, a major angiogenic factor in NSCLC [35]. The VEGF mRNA expression levels of the PSF cells were not significantly different from those of the control cell lines or clones (Figure 1C and Table 2), indicating that activation of autocrine HGF-Met loop in H358 cells did not result in an upregulation of the VEGF expression.

Tumor Histology

The xenograft tumors formed by the parent H358 cells showed sheets of poorly differentiated tumor cells with abundant eosinophilic cytoplasm. Very focally, the tumor cells formed glandular structures with small lumens, thus revealing their adeno differentiation (Figure 4B). The tumor also showed areas of necrosis especially at their center. By contrast, all tumors formed by the PSF cells lacked necrosis but formed enlarging cysts that contained serous fluid. Western blot analysis of the aspirated fluid demonstrated the presence of HGF (data not shown). All PSF tumors revealed similar histologic appearance. Focally the tumour cells demonstrated prominent papillary or acinar gland arrangements (Figure 4C). In most solid areas and that bordering the cyst lumen, the tumour cells were flattened and stratified. Many of the cells also showed cytoplasmic vacuoles that stained positively with periodic acid Schiff but were diastase resistant (PASD) (Figure 4D). Overall there appears to be a greater glandular cell differentiation with increased secretory function and putative neutral (PASD positive) mucin production.

Discussion

We have demonstrated that an overexpression of HGF in a human lung adenocarcinoma cell line that expressed its receptor Met increased the capacity for anchorage-independent growth and tumour formation in immune-deficient mice. Although HGF overexpression has been reported to increase tumorigenicity of several rodent and other human cell lines, our data represent the first direct evidence for the role of autocrine HGF-Met signaling in the biology of human lung cancer cells. The results provide a critical experimental support for evidence provided by Siegfried et al. [22] that lung cancer patients with high levels of HGF content have poorer prognosis than those with low HGF levels. Takanami et al. [21] have also reported that the co-expression of immunoreactive HGF and Met proteins in primary lung adenocarcinomas correlated with poorer overall survival for the patients.

In general, HGF-Met signaling has been considered to function mainly as a paracrine mediator of mesenchymal-epithelial cellular interaction or communication. There is, however, a large body of evidence to indicate that lung cancer especially adenocarcinoma commonly co-expresses HGF and Met. Takanami et al. [21] reported the presence of significant immunoreactive HGF-like protein in tumour cells of 40% (43 of 110) lung adenocarcinoma, and 56% of them also showed high levels of Met immunoreactivity. Harvey et al. [19] reported positive HGF immunostaining in tumour cells of 73% of non-small-cell lung cancer. In our own study, 52% (32 of 61) of primary adenocarcinoma showed strong or diffuse positive immunohistochemical staining for both HGF and Met proteins, and in situ hybridization confirmed that the HGF mRNA was expressed primarily in the tumour cells (unpublished data). It is worth noting here that the mean content of HGF in primary lung cancer tissues as assayed by Western blot technique was 27.2 ng/40 µg of tumor protein (22), with almost one fifth of tumors containing greater than 50 ng (per 40 µg tumor protein extract) HGF. Thus, the levels of HGF we detected in the xenograft tumor tissues may be too deviant from the natural in vivo situation.

The enhancement of tumorigenicity by transfection of HGF gene into human cells has been reported in breast cancer and glioma cell lines [18,36], but the effect in these models appear to be mediated by the angiogenic activity of HGF [9,37]. HGF may also enhance the expression of VEGF, hence potentiating further its own angiogenic effect [38–40]. The oncogenic effect of HGF-Met autocrine signaling activity has been demonstrated mainly in mesenchymal cells (NIH3T3, SK-LMS-1, C127) [15], or in murine carcinoma (NBT-II) cells [16]. Our findings indicated that autocrine mechanism is most likely responsible for the enhancement of the tumorigenicity of H358-PSF cells, because these cells also demonstrated increased anchorage-independent growth capacity that is independent of angiogenesis. Despite the lack of increase in VEGF mRNA expression in PSF cells, however, a contribution by the paracrine angiogenic activity of HGF cannot be completely excluded. Hence, the molecular mechanism that mediates increased anchorage-independent growth and tumorigenicity in HGF overexpressing H358 cells remains to be investigated. The oncogenicity of HGF-Met autocrine signaling in other cell lines has been associated with the activation of urokinase plasminogen activator (uPAR) [41,42], activation of α_vβ₃ integrin [43], and decreased expression of fibronectin [44].

HGF can induce clusters of epithelial cells that are growing in collagen gel to form branching tubular structures [8], a phenotype that can be equated with glandular cell differentiation. HGF-Met interaction has been shown to play important roles in the differentiation of epithelial organs including lung during normal development [10,45]. We have also provided the first evidence that autocrine HGF-Met signaling may induce in vivo differentiation in lung cancer cells. This result is consistent with a previous report by Brinkmann et al. [46], which showed induction of morphogenic activity in vitro (alveolar-like lumen formation) by HGF in two lung cancer cell lines.

In conclusion, we have provided important and direct evidence that the activation of autocrine HGF-Met signaling in lung adenocarcinoma cells may enhance tumorigenicity and differentiation. The signal transduction mechanisms and molecules that mediate these autocrine HGF-Met consequences remain to be investigated.

Abbreviations

HGF

hepatocyte growth factor

SF

scatter factor

NSCLC

non-small-cell lung carcinoma

VEGF

vascular endothelial growth factor