Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Lung cancer is the most common cause of death from malignancy in the United States and is occurring in epidemic proportions worldwide. There are nearly 200,000 newly diagnosed cases of lung cancer in the U.S. per year, with an overall 5-yr survival rate around 13% (1). Although improvements in treatment have been made, new approaches are needed to improve long-term survival. Growth factors and their receptors that are important in the growth and progression of lung tumors are potential targets for lung cancer therapy.

Hepatocyte growth factor (HGF) was first described as a mitogen and motogen for mature hepatocytes, and subsequently found to be identical to scatter factor (2, 3). HGF has recently been characterized as a multipotent growth factor that has diverse effects on a wide variety of cells in many organs (lung, kidney, breast, liver, and gastrointestinal tract). In a number of cell-culture systems, HGF has been shown to act as a mitogen, motogen, and morphogen, and an angiogenic agent, and to promote tumor progression and invasion (4). HGF is produced by mesenchymal cells, particularly fibroblasts, and acts on epithelial and endothelial cells in a paracrine fashion through binding its receptor, the c-Met protein (5). HGF is a mitogen for both normal and neoplastic bronchial epithelial cells (6), and elevated levels of HGF have been reported in various inflammatory lung diseases (7). Moreover, HGF has been shown to be a negative prognostic indicator of survival in both breast cancer (8) and non-small cell lung cancer (NSCLC) (9). Collectively, these data suggest that HGF is an important regulator of both normal and neoplastic cell growth and epithelial wound healing in the lung.

Current laboratory models of lung cancer use rodent systems and human cells injected subcutaneously or intravenously into immunodeficient mice (10). Although cells injected into the subcutaneous tissue of immunocompromised mice permit studies of cell growth in an in vivo milieu, a more representative model for studies of lung tumor cell biology would allow human tumors to be grown on a human lung substratum. Our finding that HGF is a negative prognostic indicator in NSCLC (9) suggests that HGF promotes the growth and progression of lung cancer. To address this in a clinically relevant laboratory system, we developed an in vivo model of human lung cancer by injecting primary human lung cancer cells into human bronchial xenografts (13). Our previous characterization of this model revealed that: (1) airway epithelium initially desquamates during the first week and subsequently regenerates to histologically normal-appearing airway epithelium beyond the third week after implantation; (2) the bronchial microvasculature remains phenotypically human with only a small number of murine vessels in the outer portion of the graft; and (3) xenografts remain histologically normal for up to 5 mo. After establishing the optimal conditions for the growth of primary lung-cancer cells in this model, we examined the effects of exogenously administered HGF on lung cancer growth.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Effects of HGF on 201T Cells In Vitro

201T cells were cultured in our laboratory from an adenocarcinoma of the lung, as previously described (14). We had previously shown by reverse-transcribed polymerase chain reaction analysis and Western blot that 201T cells express c-met, the receptor for HGF, at levels comparable to other lung tumor cell lines that proliferate in the presence of HGF (6). To determine the responsiveness of 201T cells to HGF, scatter assays and colony assays were performed using established protocols (6, 15).

Creation of Human Airway Tumor Xenografts

Human bronchial xenografts were created using airway segments obtained from native lungs of patients undergoing lung transplantation for end-stage lung diseases, including emphysema, pulmonary fibrosis, and pulmonary vascular diseases. Use of human tissue was approved by the University of Pittsburgh Institutional Review Board. The third- to sixth-generation bronchi, with an internal diameter of 1 to 4 mm and length of 1 to 2 cm, were dissected from surrounding lung parenchyma, ligated to silicone tubing, and implanted subcutaneously into the flanks of Severe Combined Immunodeficient (SCID) mice (16), as previously described (13). To develop a xenograft model of human lung cancer, we injected 10⁶ primary lung-cancer cells that were cultured from an adenocarcinoma of the lung (designated 201T cells), as previously described (14). Briefly, 201T cells were grown on tissue-culture plastic to 80% confluency, trypsinized to a single cell suspension, and washed in Eagle's minimum essential medium (EMEM)/N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid (HEPES) containing 10% fetal bovine serum. Cells were resuspended at a concentration of 10⁶ cells/100 µl of media and 100 µl were injected into the lumen of the xenograft after flushing with normal saline. In preliminary studies, cells were injected at varying time points after creation of the airway xenograft to determine the efficiency of tumor establishment. To determine the growth pattern of the tumor foci, xenografts were harvested at various time points and subjected to histologic analysis. To accomplish this, the airway tumor xenografts were removed from the flank of euthanized mice, and 1- to 2-mm rings of airway segment were fixed in 10% formalin. Rings were embedded in paraffin and stained with hematoxylin and eosin (H&E) for analysis.

Effects of HGF on Tumor Growth

To determine the effects of HGF on growth of 201T cells, xenografts were created on Day 0, and on Day 7 ~ 1 million 201T cells were injected into the xenograft lumen in 100 µl of EMEM/HEPES. On Days 14 through 24, 100 ng of recombinant HGF (1 µg/ml) in 100 µl was instilled into the lumen of one xenograft and the contralateral xenograft received 100 µl of sterile saline. On Days 35 through 40, xenografts were harvested, serially sectioned, and either fixed in 10% formalin or frozen in OCT compound for histologic analysis.

Histology and Immunocytochemistry

Sections, 6 µm, of paraffin-embedded tissue were deparaffinized with xylene and stained with H&E or silver using standard histology procedures. Serial sections were also immunostained using an immunoperoxidase method. Briefly, slides were incubated in 5% goat serum to block nonspecific binding of the secondary antibody. After washing, the slides were incubated with a primary antibody against human MUC1 (4H5 [17]) and MUC1 Core (Novacastra Laboratories, Ltd., Newcastle upon Tyne, UK) for 1 h at room temperature in a humidified chamber. Slides were washed in phosphate-buffered saline and incubated with secondary antibody (biotinylated goat antimouse; Vector Labs, Burlingame, CA) for 30 min. 3-Amino-9-ethylcarbazole was used to detect secondary antibody. Negative controls included omission of the primary antibody and use of an irrelevant primary antibody.

Analysis of Tumor Location and Volume

In each individual xenograft, sections from serial rings were evaluated for tumor growth and invasion. To accomplish this, H&E and immunostained sections at ~ 2-mm intervals were examined for intraepithelial and submucosal tumor. Tumor cells on the luminal side of the basement membrane were defined as intraepithelial, and cells beneath the basement membrane defined as submucosal. Within each area, tumor cell number was assessed using the following scale from 0 to 4: 0 = no detectable tumor cells, 1 = < 5 single cancer cells, 2 = multiple small (< 5 cells) foci of tumor cells, 3 = > 25% total intraepithelial or submucosal area contains tumor, 4 = > 50% total intraepithelial or submucosal volume contains tumor. This grading system allowed for comparison of intraepithelial and submucosal tumor volume between control and HGF-treated xenografts. To minimize the potential for sampling error, and because of heterogeneity within xenografts, each section was graded and the score for each xenograft assigned on the basis of the maximal tumor-cell number among the sections from each xenograft. Statistical analyses were performed using the Wilcoxon Signed Rank Test (Statview software; Abacus Concepts, Inc., Berkeley, CA).

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Response of 201T Cells to HGF In Vitro

To determine the responsiveness of 201T cells to recombinant HGF, scatter assays were performed as previously described (15). Incubation of 201T cells with 35 or 50 ng/ml of HGF induced a scatter response. Tight colonies that are normally observed with 201T cells developed pseudopodia and scattered into isolated cells in response to HGF, as shown in Figure 1. The scatter response was comparable to that observed with control Madin-Darby canine kidney cells (not shown). Colony assays showed that HGF at 10 and 35 ng/ml caused a significant increase in colony formation, as shown in Figure 2. The mean number of colonies observed was 32 colonies/well at 10 ng/ml HGF and 25 colonies/well at 35 ng/ml, compared with 19 colonies/well in the saline control (P < 0.05 by analysis of variance [ANOVA]). Collectively, the scatter and colony assays demonstrate that the 201T cells respond in vitro to HGF similar to other lung-cancer cell lines that have been shown to express the receptor for HGF, c-met (6).

View larger version (116K): [in this window] [in a new window]   Figure 1.   Scatter assay using 201T cells. 201T cells were exposed to vehicle or 10 ng/ml recombinant HGF for 18 h. As shown in A, cells exposed to vehicle are closely attached to one another. In contrast, cells exposed to HGF have separated from each other and display elongated pseudopodia (B).

View larger version (12K): [in this window] [in a new window]   Figure 2.   Colony assay. 201T cells were exposed to vehicle or 5, 10, or 35 ng/ml of recombinant HGF for 7 d, and colony numbers were determined by phase microscopy. Shown is the mean +/ standard error of the mean number of 201T cell colonies. Cells incubated with 10 or 35 ng/ml had significantly more colonies (*P < 0.05 by ANOVA).

Characterization of Lung Cancer Xenograft Model

In preliminary studies we determined the optimal conditions for creation of tumor foci in the xenografts. When cells were injected into the lumen of a mature xenograft, that is, a fully re-epithelialized xenograft 4 wk after implantation of the airway segment (13), no tumor cells could be identified either 4 or 8 wk later (data not shown). In contrast, when 201T cells were injected 7 to 10 d after creation of the xenograft, intraepithelial foci of cancer cells were present within areas of differentiated airway epithelium 3 to 5 wk later, as described later.

Because the goal of our studies was to develop an in vivo model for lung-cancer invasion of airway stroma, we initially examined the integrity of the epithelial basement membrane at the time of tumor-cell injection. To accomplish this, xenografts were harvested 7 to 10 d after creation, and then silver-stained sections were examined. As shown in Figure 3, areas of epithelial sloughing are interspersed between areas of undifferentiated airway epithelium. Silver staining revealed the basement membrane to be intact at the luminal edge of the xenograft ring, but more prominent beneath areas of undifferentiated epithelium. This suggests that the xenografted airway segments have an intact basement membrane at the time of tumor-cell injection, and that tumor cells do not have direct access to the submucosa.

View larger version (123K): [in this window] [in a new window]   Figure 3.   Basement membrane integrity in airway xenografts. Human airway xenografts were created by implanting bronchial segments into the flanks of SCID mice. At 7 d later, xenografts were harvested and examined histologically. Silver-stained sections were examined to determine the state of the epithelial basement membrane. A and B are representative sections from three xenografts. Note basement membrane beneath undifferentiated epithelium (arrowheads), and less-intense silver staining of basement membrane in area of denuded epithelium (arrows in A). Original magnification: A, ×100; B, ×400.

To determine the growth of 201T cells in airway xenografts, 201T cells were injected 7 to 10 d after creation of the airway xenografts, and harvested 2, 5, 9, or 14 wk later. At 2 wk after injection of 201T cells, there were isolated islands of adenocarcinoma within the airway epithelium (not shown). At 5 wk after injection, as shown in Figures 4A and 4B, there were isolated foci of adenocarcinoma with nodular growth into the airway lumen. At 9 wk after tumor-cell injection, there was near complete filling of the xenograft lumen by tumor foci (see Figure 4C), typically with a single identifiable stalk connecting the luminal focus to the basement membrane of the airway wall (see Figures 4C and 4D). At 14 wk after tumor-cell injection, the xenograft lumen was completely filled with adenocarcinoma and the normal bronchial epithelium was almost entirely replaced by adenocarcinoma adherent to the airway wall (see Figures 4E and 4F). There were rare areas of penetration through the basement membrane, as seen in Figure 4D, however there were no areas of tumor cell invasion through the bronchial wall of the airway xenograft.

View larger version (177K): [in this window] [in a new window]   Figure 4.   Growth of 201T adenocarcinoma in human airway xenografts. A total of 1 million primary human adenocarcinoma cells were injected into human airway xenografts 7 d after implantation into the flanks of SCID mice. The xenografts were harvested at varying time points, and serial rings examined for tumor foci. In each panel, arrowheads indicate normal epithelium and arrows indicate areas of lung cancer. (A) At 5 wk after tumor-cell injection, there were foci of adenocarcinoma in the normal airway epithelium, with extension into the airway lumen (arrow). (B) On higher magnification, tumor cells are adherent to the basement membrane (arrows) adjacent to normal differentiated bronchial epithelium (arrowheads). (C) At 9 wk after injection, the airway lumen is nearly filled with adenocarcinoma. (D) Higher magnification reveals tumor cells adherent to the airway stroma and extending beneath the basement membrane of normal epithelium in a single area (arrow). (E) At 14 wk after injection, the airway lumen is filled with adenocarcinoma and most of the normal bronchial epithelium is obliterated. (F) Tumor cells adherent to the airway wall are visible, but there is no significant invasion into the submucosa. Original magnification: A, C, and E: ×40; B, D, F: ×200. In C and E, c indicates bronchial cartilage.

These results demonstrate that injection of 201T adenocarcinoma cells into the lumen of regenerating human airway xenografts leads to the creation of tumor foci in the epithelium. Over a period of 14 wk there was growth into, and complete filling of, the airway lumen. These results were used to define the optimal conditions for creation of tumor xenografts and the experimental protocol to determine the effects of recombinant HGF on adenocarcinoma growth within the airway in this novel in vivo model. Because we wished to determine the effects of HGF on the growth of early lung cancer, we used the following protocol. On Day 0, 7 to 10 d after creation of airway xenografts, 201T cells were injected. From Days 10 through 20, HGF or vehicle was injected daily into the airway lumen. Xenografts were harvested on Days 35 to 40, at 5 wk after injection of the 201T cells.

Immunostaining to Increase Sensitivity of Detection of Adenocarcinoma in Airway Xenografts

Our preliminary studies indicated that light microscopy alone disclosed areas of 201T adenocarcinoma in the airway epithelium and lumen. We were concerned, however, that routine light microscopic analysis could fail to reveal isolated tumor foci, both in the surface epithelium and the submucosa. To increase the sensitivity of detection of small numbers of adenocarcinoma cells among normal bronchial epithelial cells and the submucosa, immunostaining for the mucin MUC1 was performed. MUC1 is transmembrane mucin expressed in the apical membrane of differentiated bronchial epithelium and in the serous cells of submucosal glands (18, and unpublished results [J. M. Pilewski]). MUC1 is also overexpressed in a large percentage of adenocarcinomas of the lung and breast (19). In adenocarcinomas, MUC1 is aberrantly glycosylated and recognized by antibodies that bind to peptide epitopes in the tandem repeat region that are normally masked by extensive O-glycosylation. We therefore used antibodies 4H5 and MUC1 Core, which recognize underglycosylated tumor MUC1 but react poorly or not at all with MUC1 expressed on normal airway epithelium (22, and unpublished results [J. M. Pilewski]), to identify 201T adenocarcinoma. As shown in Figure 5, immunostaining revealed foci of 201T adenocarcinoma in the airway epithelium (Figure 5A) and submucosa (Figure 5B) of tumor xenografts.

View larger version (62K): [in this window] [in a new window]   Figure 5.   Immunohis-tochemistry for detection of 201T cells. Airway xenografts were injected with 201T adenocarcinoma cells and 2 wk later saline or HGF was injected to contralateral xenografts as described in MATERIALS AND METHODS. Representative frozen sections from an HGF injected xenograft are shown. Note immunoreactive cells in the superficial epithelium within histologically normal epithelium (arrowheads in A) and foci of immunoreactive cells in the airway submucosa (arrows in B). Original magnification: A, ×200; B, ×100.

Effect of HGF on Intraepithelial and Submucosal Growth of 201T Adenocarcinoma

A total of 18 xenografts (nine HGF and nine saline controls) in four separate experimental groups were analyzed for tumor volume and submucosal invasion. There was a significantly greater volume of both intraepithelial and submucosal adenocarcinoma in the xenografts that were injected with HGF. Figure 6 shows representative sections of control (Figures 6A and 6C) and HGF-treated (Figures 6B and 6D) xenografts. The control (saline-treated) side displays an intact surface epithelium without gross tumor (Figure 6A). High-power inspection of the same xenograft (Figure 6C), shows a small intraepithelial focus of tumor cells (Figure 6C, arrow). The HGF-treated xenografts (Figures 6B and 6D) show foci of adenocarcinoma (arrows) in the bronchial submucosa.

View larger version (139K): [in this window] [in a new window]   Figure 6.   Histology of control and HGF-injected human airway tumor xenografts. Tumor xenografts were created and injected with saline or HGF as described in MATERIALS AND METHODS. Representative sections from serial rings are shown from saline- (A and C) and HGF-injected (B and D) xenografts. Normal bronchial epithelium is indicated by arrowheads; foci of intraepithelial and submucosal adenocarcinoma are indicated by arrows. Original magnification: A, ×40; B and D, ×100; C, ×200.

Morphologic analysis of H&E-stained and immunostained xenografts was performed to determine the effects of HGF on adenocarcinoma growth and invasion. Using the grading system described in MATERIALS AND METHODS, multiple sections from each individual xenograft were carefully evaluated for relative volume of intraepithelial and submucosal tumor and assigned a numerical grade (0 to 4). Figure 7 summarizes the results of the morphologic analyses. There was a statistically significant difference in intraepithelial tumor volume between the control and the HGF-treated xenografts (see Figure 7A). Seven of nine control xenografts had small foci of intraepithelial tumor. In contrast, all the HGF-treated xenografts had identifiable intraepithelial foci of 201T adenocarcinoma, and the relative volume was significantly increased (P = 0.042 by Wilcoxon signed rank test). Similarly, the HGF-treated xenografts had a significantly greater amount of submucosal 201T adenocarcinoma. As shown in Figure 7B, tumor cells could be identified histologically in the submucosa of only one of nine saline-treated xenografts. In contrast, foci of 201T adenocarcinoma were present in the submucosa of six of nine HGF-treated xenografts (P = 0.05 compared with control by Fisher's Exact Test), and the relative tumor volume was significantly greater in HGF-treated xenografts compared with control (P = 0.043 by Wilcoxon signed rank test).

View larger version (13K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   Figure 7.   Effects of HGF on intraepithelial and submucosal tumor volume. Tumor xenografts were created and HGF or saline was injected as described in MATERIALS AND METHODS. Serial rings were sectioned and tumor growth was evaluated using light microscopy and immunostaining for MUC1. The intraepithelial (A) and submucosal (B) tumor volumes were assessed using a relative scale, as described in MATERIALS AND METHODS. Shown are the results from nine saline- and nine HGF-injected xenografts that were processed in parallel. The horizontal line indicates the mean score for each group. Means are significantly different between control and HGF-treated xenografts (*P < 0.05 by Wilcoxon signed rank test).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

These studies demonstrate the development of a novel in vivo model of human NSCLC and reveal that HGF promotes both intraepithelial and submucosal growth of human lung tumors. By injecting human primary lung tumor cells into heterotopically transplanted human airway segments in SCID mice, an in vivo model of human lung cancer was established to study the growth of human lung tumors within the human airway. This model closely mimics the natural history of human lung tumors that develop in the bronchi or bronchioles by invading the basement membrane of surface epithelium and growing into the submucosal spaces, where the tumor can gain access to the vasculature. 201T cells, derived from a human lung adenocarcinoma, were introduced into the lumen of the airway xenografts. The cells implanted into the surface epithelium within 2 wk and formed small tumorlets that were visualized by light microscopy and immunostaining with a tumor-specific marker, in this case using an antibody specific for the underglycosylated form of the mucin MUC1. This antibody reacted only with 201T tumor cells in the xenografts, not with the normal surface or glandular epithelium. 201T cells have shown a modest ability to penetrate the basement membrane in the denuded rat trachea model (Klein-Szanto, personal communication). This result was confirmed in the human airway xenografts, inasmuch as 201T cells also showed a limited ability to invade the human airway basement membrane within 6 wk; only one of nine control xenografts contained submucosal tumor. Moreover, in the time-course experiments there was minimal invasion of the submucosa at 14 wk after 201T cell injection.

This airway tumor xenograft model differs from the denuded rat trachea model developed in the 1970s and used successfully for studies of carcinogenesis (11) and tumor invasion (12). In the denuded rat trachea model, human lung cancer cells are injected into a rat tracheal segment that has been denuded by repetitive freezing and thawing. The tracheal segment is then implanted subcutaneously into the flank of a nude mouse. When human tumor cells are injected, some of the cells adhere to the residual rat tracheal stroma and, depending on the tumor type, invade the submucosa. Although there is some revascularization of the tracheal segment by subcutaneous murine endothelial cells, the stroma consists primarily of residual devitalized rat trachea. The human airway tumor xenograft model developed by us is an improvement over the denuded rat tracheal model insofar as the surrounding human epithelium, stroma, and microvasculature of the xenograft remain intact, thereby more closely approximating the milieu of native human lung cancers. Other advantages of this model include the ability to intraluminally administer recombinant growth factors, and the ability to perform immunocytochemistry with murine-derived antibodies.

The human airway tumor xenograft model has some similarities to a recently described in vitro model of lung cancer invasion. Al-Batran and colleagues (23) cocultivated bronchial biopsies with lung cancer cells, and reported persistence of normal epithelium and invasion of lung cancer cells over 6 wk. Exposure of lung cancer cells to normally differentiated epithelium resulted in inefficient tumor cell attachment, whereas exposure of lung cancer cells to injured epithelium resulted in reproducible tumor attachment. Our studies confirmed this observation, as tumor foci developed in airway xenografts only when introduced during the period of epithelial shedding and regeneration (Days 7 to 10) that results from airway ischemia. In addition to apical membrane barriers to cell adhesion (tight junctions and cell polarization that restrict access to matrix proteins), secreted and transmembranous mucins present in the xenograft model likely contribute to the inefficient tumor cell attachment to normal airway epithelium. Notably, in our model, the luminal basement membrane, as assessed by silver staining, was intact at the time of tumor-cell injection. This suggests that submucosal extension required tumor cell-mediated invasion of the basement membrane.

These experiments also demonstrate that tumor invasion and growth within airway xenografts can be modulated by HGF, a growth factor that is known to stimulate cell motility and cell proliferation in vitro. Daily injection of recombinant HGF into the tumor xenografts induced extensive invasion and accelerated growth of 201T cells compared with control xenografts (six of nine HGF-injected xenografts, versus one of nine control xenografts, had submucosal tumor). A possible mechanism for the effects of HGF is that, through activation of its tyrosine kinase signaling pathway, HGF stimulates transcription of genes for metalloproteinases that enhance breakdown of basement membrane (24). Such proteases include matrix metalloproteinase (MMP)-9 (24), type IV collagenase (MMP-2 [26]), and urokinase-type plasminogen activator (25). An alternative is that the observed effects of HGF on lung tumor cell growth were an indirect result of HGF affecting the bronchial epithelium. The lack of significant histologic change in the bronchial epithelium speaks against this possibility. Further studies with an HGF antagonist will be necessary to prove that tumor growth and invasion is due directly to HGF.

Our in vivo results confirm our previous data from in vitro experiments that demonstrated the ability of HGF to cause human lung tumor cells to migrate through nylon membranes and to invade artificial extracellular matrix (6). HGF had no noticeable effect on the normal surface or glandular epithelium within the xenografts, but significantly increased both the number of xenografts containing submucosal tumor and the size of the tumors. Our model suggests that HGF acts on human lung tumors through stimulation of both proliferation and invasive behavior. Microenvironments high in HGF may encourage aggressive tumor growth in a paracrine fashion; this has been reported in an in vitro coculture model (26). This coculture study, as well as our current study, confirms the observations that HGF is associated with aggressive tumor behavior and poor patient outcome in both breast and lung cancer (8, 9). The human airway xenograft model provides an excellent system for the study of inhibitors of lung-tumor invasion, drug or gene delivery to lung tumors established within the human lung epithelia, and expression of enzymes such as metalloproteinases whose expression may be enhanced during the invasive process.