Five-week-old female nude mice (BALB/c nu/nu; Oriental Yeast Ltd., Tokyo, Japan) were used as experimental animals. Food was supplied ad libitum and the animals were housed under sterile conditions. All animal experiments were performed in compliance with the Guidelines for Experimental Animals of Hyogo College of Medicine.

N^G-nitro-L-arginine-methyl ester (L-NAME), a NOS inhibitor and 1,1′-{1,4-phenylenebis(methylene)}bis-1,4,8,11-tetraazacyclotetradecane octahydrochloride (AMD3100), a CXCR4 antagonist, were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA). Dihydrochloride (1400W), a selective iNOS inhibitor, was purchased from Cayman Chemical Co. (Ann Arbor, MI, USA). These agents were dissolved in Mg²⁺ and Ca²⁺-free phosphate-buffered saline (PBS[−]) before use.

The original tumor tissue was obtained from the surgical specimens of a 72-year-old man with ACC of the oral floor. Histologically, the tumor showed a cribriform pattern. Metastatic tumor tissue was obtained from a lesion of spontaneous neck metastasis from the above ACC transplanted into a second passage of nude mice. The tumor was rinsed 3 times in PBS, cut into ~2×2-mm pieces and transplanted into the flanks of the mice. At the second passage level, ~1 year after ACC transplantation, another tumor mass appeared in the neck, apart from the transplanted site of the tissue fragment. Although this tumor was an undifferentiated carcinoma histologically, it was considered a metastatic lesion and designated as ACCIM. ACCIM produced multiple metastases to lymph nodes and lungs 5 months after transplantation (24).

ACCIM was cut into ~2×2-mm pieces and implanted subcutaneously in nude mice. Approximately 19 days after ACCIM xenotransplantation, when the tumor reached 10 mm in diameter, mice were subdivided into six groups and were assigned to receive one of the following treatments by intraperitoneal injection every day for 5 weeks: a) vehicle (PBS[−]), b) AMD3100 (2 mg/kg), c) L-NAME (20 mg/kg), d) 1400W (5 mg/kg), e) both AMD3100 and L-NAME (AMD3100+L-NAME), or f) both AMD3100 and 1400W (AMD3100+1400W). Five weeks and 28 weeks after the start of treatment, 5 and 4 mice from each group were randomly selected and sacrificed, respectively (Fig. 1). The tumor volume was calculated by the following formula: volume (mm³) = a²xb/2, where a is the tumor width in mm and b is the tumor length in mm (25). Necropsies were performed to identify macro-metastases to the lung 28 weeks after treatment began. The primary tumors and lungs were harvested, fixed in 10% formalin, embedded in paraffin, cut into 4-μm-thick sections and stained with hematoxylin and eosin (H&E) according to conventional procedures.

Immunohistochemical examination was performed with the use of the avidin-biotin-peroxidase complex (ABC) staining method (26). Briefly, endogenous peroxidase activity was blocked in tissue specimens by treatment with 0.3% H₂O₂ in methanol for 5 min. The specimens were washed and treated with 1% normal horse serum in PBS for 15 min. After washing with PBS, rabbit polyclonal antibody for human CXCR4 (Abcam, Cambridge, UK), rabbit polyclonal antibody for human NOS3 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), or rabbit polyclonal antibody for human NOS2 (Santa Cruz), was applied as primary antibody at 4°C overnight. After further washing with PBS, the specimens were incubated with ABC complex solution (Vectastain, Vector Lab., Burlingame, CA, USA) at room temperature for 15 min. After washing with PBS, biotinylated goat anti-mouse IgG (Vector) was applied to the sections, which were then incubated for 30 min at room temperature. The specimens were treated for ~5 min with a substrate solution containing 3,3′-diaminobenzidine tetrahydrochloride (Wako Pure Chemical Industries, Ltd., Osaka, Japan) and H₂O₂. Finally, the specimens were counterstained with hematoxylin, dehydrated and mounted with glycerol gelatin.

The CXCR4, iNOS and eNOS labeling indexes (LI) were obtained by calculating the ratio of positive cells to the total number of tumor cells counted in well-labeled areas, as determined by scanning twenty areas at ×200 magnification.

Tumor samples were lysed in a lysis buffer consisting of PBS[−], supplemented with 20 mM Tris-HCl, pH 8.0, 1% NP40, 150 mM NaCl, 1 mM EDTA, 10% glycerol, 0.1% β-mercaptoethanol, 0.5 mM dithiothreitol and a mixture of proteinase inhibitors consisting of 1 mM phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 5 μg/ml leupeptin, 5 mM benzamidine, 1 μg/ml pepstatin, 2 μg/ml antipain hydrochloride (Boehringer, Mannheim, Germany), 50 μM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride (Wako Pure Chemical Industries), 2 mM sodium orthovanadate (Sigma-Aldrich) and 20 U/ml ulinastatin (Mochida Pharmaceutical, Tokyo, Japan). Lysates containing 15 μg protein were subjected to electrophoresis in a 10–20% gradient SDS-PAGE mini gel (Bio-Rad, Chicago, IL, USA) and blotted onto a PVDF membrane using Multiphor II (Amersham Pharmacia Biotech, Buchinghamshire, UK) for 30 min. The blotted membrane was blocked with 5% skim milk in 10 mM Tris-HCl, pH 7.2, containing 150 mM NaCl and 0.5% Tween-20 and was incubated with primary antibodies (0.1–1 μg/ml) at 4°C for 16 h as described below. The membrane was then incubated with alkaline phosphatase-conjugated secondary antibodies (0.02 μg/ml) for 4 h at room temperature as described below. The membrane was rinsed and then treated with nitroblue tetrazolium (Sigma-Aldrich) and 5-bromo-4-chloro-3-indolyl phosphate (Sigma-Aldrich) to visualize the protein bands. The primary antibodies used were rabbit polyclonal antibody for human CXCR4 (Abcam) and rabbit polyclonal antibody against iNOS and eNOS (Santa Cruz). The secondary antibodies used were anti-rabbit IgGs conjugated with alkaline phosphatase (Santa Cruz). Actin was used as an internal control.

To detect DNA breaks, in situ terminal doxynucleotidyl transferase (TdT)-mediated dUTP-digoxigenin nick-end labeling (TUNEL) was performed as described by Gavroieli et al(27). Briefly, after deparaffinization and blocking of endogenous peroxidase with 0.3% H₂O₂ in methanol for 30 min at room temperature, the sections were treated with 20 μg/ml proteinase K (Dako Cytomention, Glostrup, Denmark) for 15 min at room temperature. The sections were submitted to TdT reaction in the presence of terminal transferase and biotin-16-dUTP for 60 min at 37°C. The sections were then incubated with diluted peroxidase conjugated streptavidin for 30 min at room temperature to detect biotin-16-dUTP labeling, followed by color development with a solution containing 3,3′-diaminobenzidine and H₂O₂. Methyl green was used for counterstaining. TUNEL-positive cancer cells were counted in twenty areas at high magnification (×400) that show the well-labeled areas.

Microvessels were detected by immunohistochemical staining for CD31, a marker for vascular endothelial cells. After pretreatment with 0.25% trypsin for 10 min, tissue sections were immunostained with an anti-human CD31 mouse monoclonal antibody (Novocastra, Newcastle Upon Tyne, UK) using the SABC method. MVD was determined by counting the number of vessels at magnification (×200) of the tumor stroma that contained the highest number of capillaries.

Statistical analysis was done with Student's t-test. Differences were considered statistically significant when the p-value was <0.05.

All agents were well tolerated by the mice, without weight loss or signs of toxicity. These agents inhibited the proliferation of the subcutaneously xenotransplanted tumors. Treatment with 1400W for 4 and 5 weeks significantly inhibited tumor growth as compared with treatment with vehicle (p<0.05) (Fig. 2A). Combined treatment (AMD3100+L-NAME and AMD3100+1400W) also significantly inhibited tumor growth as compared with vehicle at the end of the observation period (p<0.05). The reduction rate in tumor growth did not differ significantly among these treatments (1400W, AMD3100+L-NAME and AMD3100+1400W). The final mean tumor volume per mouse was 2234±556 mm³ (54.5% decrease) in 1400W-treated mice, 2554±612 mm³ (48.0% decrease) in AMD3100+L-NAME-treated mice and 2443±602 mm³ (50.2% decrease) in AMD3100+1400W-treated mice, as compared with 4910 mm³ in vehicle-treated mice (Fig. 2B and C). These tumors were undifferentiated carcinoma histologically (Fig. 2C).

The immunoreactivity of each specimen was evaluated by light and transmission microscopy to assess the intensities of CXCR4, iNOS and eNOS expression. The labeling index (LI) of CXCR4 was significantly lower in tumors treated with AMD3100 (59.8%) than in tumors treated with vehicle (89.3%). In contrast, there were no significantly differences in the LI of CXCR4 in tumors treated with L-NAME or 1400W as compared with tumors treated with vehicle (Fig. 3). The LI of iNOS was 83.1% in vehicle-treated tumors, 79.5% in L-NAME-treated tumors, 58.0% in 1400W-treated tumors and 61.5% in AMD3100-treated tumors. The LI of iNOS was significantly lower in tumors treated with 1400W or with AMD3100 than in those treated with vehicle (Fig. 4). The LI of eNOS was 80.1% in vehicle-treated tumors, 64.5% in L-NAME-treated tumors, 54.7% in 1400W-treated tumors and 75.3% in AMD3100-treated tumors. The LI of eNOS was significantly lower in L-NAME- or 1400W-treated tumors than in vehicle-treated tumors (Fig. 5). The decreases in the LI of iNOS and eNOS but not CXCR4 in tumors subjected to combined treatment (AMD3100+L-NAME, AMD3100+1400W) were significantly greater than those in L-NAME-, 1400W- or AMD3100-treated tumors (Figs. 3–5).

The expression levels of CXCR4, iNOS and eNOS in vehicle-treated tumors on western blot analysis were compared with those in tumors treated with each antagonist and inhibitor. CXCR4 expression decreased in tumors treated with 1400W, AMD3100+L-NAME, or AMD3100+1400W. The decrease in iNOS expression was greatest in tumors treated with 1400W. However, the expression of eNOS did not differ significantly among the treatment groups (Fig. 6).

When apoptosis induction in tumor parenchyma was examined by the TUNEL method, apoptotic cancer cells identified by brown nuclear TUNEL signals were clearly observed in tumors treated with 1400W, AMD3100+L-NAME, or AMD3100+1400W. The mean apoptosis index was 0.9% in tumors treated with vehicle, but significantly increased in tumors treated with 1400W, AMD3100+L-NAME, or AMD3100+1400W (Fig. 7).

When the effects of treatment with agents on tumor-induced angiogenesis in tumor stroma were examined histologically, the MVD was significantly lower in tumors treated with 1400W, AMD3100+L-NAME, or AMD3100+1400W than in vehicle-treated tumors. There was no significant difference in the MVD between 1400W-treated tumors and tumors treated with AMD3100+L-NAME or AMD3100+1400W (Fig. 8).

In vehicle- and L-NAME-treated mice, metastasis to the lung occurred in all 4 mice (100%). After treatment with AMD3100, 1400W, AMD3100+L-NAME, or AMD3100+1400W, no lung metastasis was found (Fig. 9A). The mean number of lung metastatic lesions per mouse was 5 in vehicle-treated mice and two in L-NAME-treated mice. The diameters of all lung metastatic lesions (20 lesions) in vehicle-treated mice was ≥1.0 mm. Only one in eight metastatic lesions was >1.0 mm in diameter in L-NAME-treated mice (Fig. 9B).