Transfection of interleukin-8 increases angiogenesis and tumorigenesis of human gastric carcinoma cells in nude mice

The growth and spread of tumour cells depends on adequate vasculature. We have previously reported that the expression of interleukin-8 (IL-8) directly correlates with the vascularity of human gastric carcinomas. To provide evidence for a causal role of IL-8 in angiogenesis and tumorigenicity of human gastric cancer, we used the lipofectin method to stably transfect the human TMK-1 gastric carcinoma cells (low endogenous IL-8) with an IL-8 expression vector or control vector. Transfection with IL-8 did not affect the proliferation of cultured cells, yet the culture supernatants of the transfected (but not control) cells stimulated proliferation of human umbilical vein endothelial cells. The IL-8-transfected and control cells were injected into the gastric wall of nude mice. IL-8-transfected cells produced rapidly growing, highly vascular neoplasms as compared to control cells. These results provide direct evidence for the role of IL-8 in the angiogenesis and tumorigenicity of human gastric carcinomas. © 1999 Cancer Research Campaign

There have been many studies attempting to isolate the molecular mediators of this process. Among the possibilities is interleukin-8 (IL-8), a member of the CXC chemokine family. This cytokine was initially shown to selectively stimulate chemotactic activity for neutrophils and lymphocytes (Matsushima et al, 1988(Matsushima et al, , 1992. More recent studies revealed that IL-8 is multifunctional: it can induce angiogenesis Strieter et al, 1992), haptotactic migration (Wang et al, 1990) and proliferation of keratinocytes and melanoma cells (Krueger et al, 1990;Schandendorf et al, 1993). Human recombinant IL-8 can induce proliferation and migration of human umbilical vein endothelial cells (HUVECs) and is potently angiogenic when implanted in the rat cornea . Furthermore, IL-8 is a known angiogenic factor for human lung carcinoma (Strieter et al, 1995), human bladder carcinoma (Tachibana et al, 1997) and human prostate carcinoma (Green et al, 1997). The CXC chemokines IL-8 and interferon-α-inducible protein (IP-10) have recently been shown to regulate angiogenic activity in idiopathic pulmonary fibrosis (Keane et al, 1997).
The expression of a wide variety of growth factors/receptors and cytokines, including epidermal growth factor (EGF) transforming growth factor (TGF-α), EGF receptor (EGF-R) (Tahara, 1990(Tahara, , 1993, IL-1 (Ito et al, 1993) and IL-6 (Ito et al, 1997) by human gastric carcinoma cells has been shown to correlate with malignant potential. Gastric carcinoma cells also produce angiogenic factors, including basic fibroblast growth factor (bFGF) (Tanimoto et al, 1991) and vascular endothelial growth factor (VEGF) (Yamamoto et al, 1998). We have recently found that surgical specimens of human gastric carcinomas overexpress IL-8 as compared to corresponding normal mucosa, and that the IL-8 mRNA level directly correlated with the vascularity of the tumours (Kitadai et al, 1998b).
There have not, however, been any studies that established a causal role for IL-8 in gastric cancer angiogenesis. The purpose of this study was, therefore, to provide evidence for the causal role of IL-8 in the growth and vascularization of human gastric cancer. We show that stably IL-8-transfected human gastric cancer cells produce angiogenic activity in culture. Subsequent to orthotopic implantation into nude mice, the cells produce rapidly growing and highly vascularized tumours.

Transfection assays and production of stable cell lines
Transfections were performed by the lipofection method (Life Technologies) with the following modifications: approximately 10 6 TMK-1 cells were plated into culture dishes (90-mm diameter) 1 day prior to transfection. On the following day, the growth medium was replaced, and 3 h later, liposomes containing 5 µg of the expression plasmid were applied to the cells and left for 6 h. After that, cell monolayers were rinsed with RPMI containing 400 µg of G418 ml -1 (selection medium). The selection medium was changed every 3 days.

DNA probes
The cDNA probes used in these analyses were a 1.3-kb PstI cDNA fragment corresponding to rat glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (Clontech, Palo Alto, CA, USA) and a 0.5kb EcoRI cDNA frgament corresponding to human IL-8 (Matsushima et al, 1988). A 0.7 kb human VEGF cDNA probe and a 0.8-kb bFGF cDNA probe were kindly provided by Dr M Shibuya (University of Tokyo, Tokyo, Japan) and Dr J A Abraham (California Biotechnology, Inc., CA, USA), respectively. Each cDNA fragment was purified by agarose gel electrophoresis, recovered using GeneClean (BIO 101, Inc., La Jolla, CA, USA), and radiolabelled using the random primer technique with 32 P-labelled deoxyribonucleotide triphosphates (Feinberg and Vogelstein, 1983).

Immunohistochemical staining
Consecutive 4-µm sections were cut from each study block. Sections were immunostained for IL-8 and CD31 (specific for mouse endothelial cells). Immunohistochemistry (IHC) was performed by the immunoperoxidase technique with minor modification (Gutman et al, 1995;Kitadai et al, 1998b). Antibodies used were a rabbit polyclonal antibody (Otsuka Pharmaceutical Co. Ltd, Tokushima, Japan) at a 1:100 dilution for IL-8, and a rat monoclonal antibody (Pharmingen, San Diego, CA, USA) for CD31. Negative controls were done using non-specific IgG as the primary antibody.

Vessel density
Vessel count was assessed by light microscopy in IHC-stained areas of the tumour containing the highest numbers of capillaries and small venules at the invasive edge (Weidner et al, 1991). Highly vascular areas were first identified by scanning tumour sections at low power (×40 and ×100). Vessel count was determined in six such areas at ×400 field (×40 objective and ×10 ocular), and the average count of the six fields was determined. Vessel lumen was not necessary for a structure to be defined as a vessel (Weidner et al, 1991).

Cell growth in vitro
Cells (5 × 10 3 ) were seeded on 24-well plates (Falcon Laboratories, McLean, VA, USA) and cultured in RPMI medium in the absence or presence of FBS. The medium was changed every 48 h. Cell number was determined in triplicate cultures.

MTT assay
HUVECs (5 × 10 3 ) were plated into multiple 38-mm 2 wells of 96well gelatin-coated plates (Falcon Laboratories, McLean, VA, USA) in cultured medium from the transfected TMK-1 cells. The cells were cultured for 2 days, when their proliferation was determined by an MTT assay (Fan et al, 1990). Fifty microlitres of MTT (40 µg ml -1 ) was added to each well, incubated for 1 h, aspirated and dissolved in dimethyl sulphoxide. The intensity of colour adduct formation was measured using an ELISA plate reader. The percentage increase in cell growth was calculated as: growth stimulation (%) = (B-A)/A × 100 where A is the A540 of the control cultures and B is the A540 of test cultures. IL-8 neutralization studies used a polyclonal rabbit anti-human IL-8 antibody (Otsuka Pharmaceutical Co. Ltd).

Growth of transfected tumour cells in nude mice
Male athymic BALB/c nude mice were obtained from Charlsriver Co. Ltd (Tokyo, Japan). The mice were maintained under specific pathogen-free conditions and used when 8 weeks old. To produce tumours, cultured cells were harvested by a brief trypsinization, and 5 × 10 5 or 1 × 10 6 viable cells were implanted into the gastric wall of nude mice as described previously (Takahashi et al, 1995). Six weeks later, the mice were killed and the tumours growing in the stomach were removed, weighed and examined histologically.

Statistical analysis
The significance of the in vitro data was analysed by the Student's t-test (two-tailed), and the in vivo data was analysed by the Mann-Whitney U-test.

Transfection of TMK-1 cells with the IL-8 expression vector
We used the TMK-1 cell line derived from poorly differentiated adenocarcinoma because its expression level of IL-8 is extremely low (Kitadai et al, 1998b). First, we transfected the IL-8 expression vector (IL-8-BCMGS-neo) into the TMK-1 gastric carcinoma cells and selected stable IL-8-overexpressing clones. ELISA demonstrated that clone 15 secreted the highest level of IL-8 protein into the culture medium (Table 1); this clone was used in subsequent experiments. The IL-8 mRNA expression and cytosol localization of IL-8 protein were confirmed by Northern blot and IHC analyses respectively (Figure 1). As a control, we used the   (Table 1 and Figure 1). We next examined whether the IL-8 protein produced by the IL-8 transfectant was biologically active. Since IL-8 has been reported to stimulate proliferation and migration of HUVECs Kitadai et al, 1998b), we determined whether culture supernatants induced proliferation of HUVECs. Proliferation of HUVECs was significantly stimulated by the addition of medium conditioned by the IL-8 transfectants as compared to that from control cells. The growth stimulatory activity was blocked by neutralizing antibodies to IL-8 ( Table 2).

Effect of IL-8 transfection on in vitro growth of gastric carcinoma cells
Since IL-8 has been reported to be an autocrine growth factor for melanoma cells (Schandendorf et al, 1993), we next analysed whether IL-8 transfection stimulates the in vitro growth of gastric carcinoma cells. Under both serum-free and serum-containing conditions, cell growth was not affected by transfection with the IL-8 gene (Figure 2). This finding agreed with our previous data showing that the addition of exogenous IL-8 did not alter cell proliferation of gastric carcinoma cell lines (Kitadai et al, 1998b).

In vivo growth of IL-8-transfected gastric carcinoma cells
Next, we injected the control and IL-8-transfected cells into nude mice. Since the organ microenvironment influences tumour growth and metastasis, the cells were injected into an orthotopic site (gastric mucosa) of nude mice (Fidler, 1990). As shown in Table 3, transfection of TMK-1 with IL-8 increased their tumorigenicity; namely, by 6 weeks after injection of 5 × 10 5 cells, IL-8transfected cells formed tumours in seven of ten nude mice, whereas control cells (TMK-1 neo) did not form any. In addition, the size of gastric tumours produced by the 1 × 10 6 IL-8-transfected cells was significantly larger than that produced by control cells (Table 3). Stable expression of IL-8 mRNA and protein in the tumour lesions were confirmed by Northern blot analysis and ELISA respectively (Figure 3). Transfection with IL-8 gene did not change the expression levels of mRNA for VEGF and bFGF ( Figure 3A). Immunohistochemical analysis confirmed the expression of IL-8 at the cell level. IL-8 immunoreactivity within the tumours localized mainly to cancer cells ( Figure 4A). Normal epithelial cells and stromal cells showed minimal IL-8 staining with IHC (data not shown). Distant haematogeneous or peritoneal dissemination was not found in any of the injected mice.

Effect of IL-8 transfection on angiogenesis of orthotopic xenografts
To determine whether the increased tumorigencity and growth of the tumours was associated with increased angiogenesis, we performed IHC against CD31 (mouse endothelial cell-specific) and counted microvessels in the orthotopic tumours. A representative IHC for CD31 is shown in Figure 4. The IL-8 transfection resulted in stimulation of angiogenic responses. Namely, tumour vessel density in the tumours produced by the IL-8 transfectants A Table 2 Growth stimulation of HUVEC by medium conditioned by IL-8transfected TMK-1 cells

DISCUSSION
IL-8 is a multifunctional cytokine originally identified as a leucocyte chemoattractant (Matsushima et al, 1988(Matsushima et al, , 1992. It can induce migration in some tumour cells (Wang et al, 1990) and has been implicated in the induction of angiogenesis in such diverse diseases as psoriasis (Schroeder and Christopher, 1986), rheumatoid arthritis (de Marco et al, 1991), idiopathic pulmonary fibrosis (Keane et al, 1997) and some malignant diseases (Folkman, 1986(Folkman, , 1990. Previous studies have demonstrated that IL-8 is expressed by lung (Strieter et al, 1995), bladder (Tachibana et al, 1997), prostate (Green et al, 1997) and head and neck squamous carcinomas (Richards et al, 1997), astrocytoma (Van Meir et al, 1992) and malignant melanoma (Schandendorf et al, 1993;Singh et al, 1994). It has been reported that IL-8 acts as an angiogenic factor in lung carcinoma Strieter et al, 1995;Arenberg et al, 1996); however, the function of IL-8 in the other carcinomas has not been clarified. Recently, we found that most gastric carcinomas express IL-8 mRNA and protein, and its level directly correlates with angiogenic activity in the tumour (Kitadai et al, 1998b).
In the present study, we performed transfection experiments to obtain direct evidence that IL-8 regulates angiogenesis in gastric carcinoma. IL-8 expression vector was stably transfected into TMK-1 cells (which express negligible levels of IL-8 mRNA and protein) (Kitadai et al, 1998b). Transfection with the IL-8 gene did not affect in vitro cell proliferation; however, enforced expression of IL-8 in TMK-1 cells increased their tumorigenic and angiogenic potential in the gastric wall of nude mouse (orthotopic site), thus providing direct evidence for the involvement of IL-8 in angiogenesis.
Although IL-8 transfection increased tumorigenicity and angiogenicity, neither control nor IL-8 transfectants produced distant metastasis. To produce metastasis, tumour cells must complete a series of sequential and selective steps that include growth, vascularization, invasion, adhesion and extravasation (Fidler, 1990). The increase in angiogenic activity by IL-8 transfection may be necessary but not sufficient to produce metastasis by TMK-1 cells.
Angiogenic factors produced by tumour cells and normal cells are critical to the formation of a vascular bed necessary to support tumour growth at the primary and metastatic sites. Because of the complex nature of the angiogenic process, it is unlikely that any one factor is responsible for angiogenesis in a particular tumour type. Within any individual tumour, there may be a dominant angiogenic factor that favours the imbalance of the positive and negative regulators to induce angiogenesis. The two most potent angiogenic molecules are VEGF and bFGF. We previously studied the expression of VEGF and bFGF as well as IL-8 in human gastric carcinoma. The expression level of bFGF is very low and is D Figure 4 Immunohistochemistry for IL-8 (A, C) and CD31 (B, D) of TMK-1 orthotopic xenografts. Note that IL-8 immunoreactivity is detected within TMK-1 IL-8-C15 cells (C) but not TMK-1 neo cells (A). The vascularity is higher in TMK-1-IL-8-C15 tumours (D) as compared with TMK-1 neo tumours (B); ×250 associated with fibrosis in the diffuse type gastric carcinoma (Tanimoto et al, 1991). On the other hand, VEGF is commonly expressed by all the gastric carcinoma cell lines and carcinoma tissues as well as normal mucosa (Yamamoto et al, 1998). Among these angiogenic molecules, IL-8 correlates best with vascularity in the tumours. In addition, we demonstrated that overexpression of IL-8 induced an angiogenic response. Therefore, IL-8 actually regulates angiogenesis in human gastric carcinomas. In this study, we examined the expression of VEGF and bFGF by IL-8 transfectants. There were no changes in the levels of mRNA for VEGF and bFGF after transfection with the IL-8 gene ( Figure 3A), indicating that these factors did not play a role in the increased angiogenic activity in this system. Smith et al demonstrated that inhibition of IL-8 using neutralizing antibody resulted in the marked attenuation of angiogenesis in bronchogenic carcinoma . To prove a possibility for therapeutic intervention for gastric carcinoma, additional studies with neutralizing IL-8 antibody should be required.
IL-8 has been reported to act as an autocrine growth factor for melanoma cells (Schandendorf et al, 1993) and the expression level of IL-8 by human melanoma cells correlates with their metastatic potential (Singh et al, 1994). Furthermore, IL-8 up-regulates collagenase type IV mRNA expression and collagenase activity by melanoma cells (Luca et al, 1997) and stimulate cell motility (Wang et al, 1990). Recent studies have shown that IL-8 receptors, IL-8RA and IL-8RB, are expressed by not only microvessel endothelial cells but also tumour cells in head and neck squamous cell carcinoma (Richards et al, 1997) and breast carcinoma (Miller et al, 1998). Therefore, IL-8 may play an important role in tumour growth and progression by both autocrine and paracrine mechanisms. It would be of great interest to elucidate whether IL-8 receptors are expressed by gastric carcinoma cells.
In conclusion, this study demonstrates that IL-8 produced by gastric carcinoma cells regulates angiogenesis. The identification of factors that correlate with angiogenesis in gastric carcinoma may provide a basis for the design of therapeutic approaches. Studies to determine if attenuation of IL-8 production by gastric carcinomas can produce therapeutic benefits are under way.