5a-D-Luc cells, a highly bone metastatic activity cell line derived from MDA-MB-231 (6), were maintained in DMEM medium supplemented with 5% FBS (JRH Biosciences, Lenexa, KS, USA) in a humidified incubator maintained at 37°C with 5% CO₂. The 5a-D-Luc cells were transfected with pZsGreen1-C1 vectors (Clontech, Mountain View, CA, USA) using Lipofectamine 2000 Transfection Reagent (Life Technologies, Carlsbad, CA, USA), and maintained in DMEM with 5% FBS and 1 mg/ml G418 (Roche, Basel, Switzerland) to obtain stable transfectants. We subsequently isolated five subclones of the transfectants stably expressing ZsGreen1 protein (5a-D-Luc-ZsGreen).

Cell proliferation assay was performed using a sulforhodamine B as described previously (11). To evaluate fluorescence and bioluminescence intensities in vitro, 5a-D-Luc-ZsGreen cells were serially diluted from 10⁵ to 10⁴ cells in 96-well plates. The fluorescence intensity was measured with a preclinical in vivo imaging system IVIS Lumina (Xenogen, Alameda, CA, USA). Then, the bioluminescence intensity (luciferase activity) was measured using Steady-Glo Luciferase Assay system (Promega, Madison, WI, USA) with the IVIS Lumina.

The animal experiments were approved by the Institutional Animal Care and Use Committee of the National Institute of Radiological Sciences, and all animal experiments were conducted in accordance with the institutional guidelines regarding animal care and handling. To evaluate fluorescence and bioluminescence intensities in vivo, female BALB/c-nu/nu mice (5-week-old; CLEA Japan, Tokyo, Japan) were subcutaneously inoculated with a 1:1 mixture of 5a-D-Luc-ZsGreen cells (1×10³, 1×10⁴, 5×10⁴ and 1×10⁵ cells) and Matrigel (Becton-Dickenson Bioscience, Bedford, MA, USA) on the dorsal side. For intracardiac injection, female nude mice (6-week-old) were anesthetized by inhalation of 3% isoflurane and injected with 5a-D-Luc or 5a-D-Luc-ZsGreen cells (1×10⁵ cells) in 100 μl PBS into the left cardiac ventricle as described previously (6).

Mice were injected with 2.5 mg/mouse D-luciferin (Wako Pure Chemical Industries, Osaka, Japan) into the tail vain, and the bioluminescence images were acquired with the IVIS Lumina under isoflurane anesthesia (3%). Mean radiance (photon/s/cm²/sr) in region of interest (ROI) drawn over bioluminescence signals was measured using Living Image software (Xenogen).

Brain, bone, and other organs with bioluminescence signals were resected from the mice that received intracardiac injection of 5a-D-Luc-ZsGreen cells at 7, 14, 21 and 28 days after injection. The resected tissues were observed with a fluorescence stereoscopic microscope MZ16F (Leica Microsystems, Wetzlar, Germany).

For the surgical procedure, a mixture of air, oxygen, and isoflurane (1.5–2%) anesthesia was given via face-mask. The head of mouse was fixed with a stereotactic frame, and cranial windows were prepared according to the ‘Seylaz-Tomita method’ (12). The cranial window was attached over the left side of the somatosensory cortex using dental cement (Luxaflow, DMG, Hamburg, Germany) and centered at 1.8 mm caudal and 2.5 mm lateral from the bregma. A custom-made metal plate was affixed to the front of the central skull (12,13). For real-time tracking imaging of tumor cells just after inoculation, mice were anesthetized with 1.5% isoflurane and was placed supine on a custom-made apparatus. Needle (30 G: Terumo, Tokyo, Japan) with thin plastic tube (catheter) was inserted into left cardiac ventricle of mice and was fixed with dental cement (Ionosit, DMG). The head of a mouse intraperitoneally administered 10 mM SR101 (8 μl/g body weight) was placed under an objective lens of two-photon laser scanning microscopy (TCS-SP5 MP, Leica Microsystems). Immediately after initiation of two-photon imaging, 1×10⁵ of 5a-D-Luc-ZsGreen cells were injected into the mouse through the catheter. The wavelength of excitation laser was 900 nm. An emission signal was separated by a beam splitter (560/10 nm) and simultaneously detected through a band-pass filter for SR101 (610/75 nm) and tumor cells (525/50 nm). A single image plane consisted of 1,024 by 1,024 pixels, and in-plane pixel-size was 0.25–0.45 μm depending on an instrumental zoom factor. The images were acquired at depth of 0.2–0.4 mm from the cortical surface. To observe the moving cancer cells and cerebral vasculature on the surface over the barrel cortex, dynamic imaging was started just after inoculation and conducted at a rate of 0.15 sec per frame for 60 sec. For temporal imaging to observe change of tumor cells and blood vessels until day 21 after injection, the mice without a catheter were imaged as described above.

After ex vivo imaging, the bone tissues were fixed in 10% neutral-buffered formalin, decalcified with EDTA treatment and embedded in paraffin. The tissue sections (1-μm thick) were deparaffinized and stained with H&E or tartrate resistant acid phosphatase (TRAP) staining kit (Wako Pure Chemical Industries). After TRAP staining, the sections were counterstained with hematoxylin. The brains were resected, embedded in frozen section compound (Leica Microsystems) and frozen for cryostat sectioning (8-μm thick). Immunofluorescence staining was conducted using rat anti-CD31 antibody (BD Pharmingen, San Diego, CA, USA) and Alexa Fluor 594-labeled anti-rat IgG (Molecular Probes, Eugene, OR, USA) or rabbit anti-claudin-5 antibody (Abcam, Cambridge, UK) and Alexa Fluor 594-labeled anti-rabbit IgG (Molecular Probes). The slides were mounted in Vectashield mounting medium with DAPI (Vector Laboratories, Burlingame, CA, USA). The stained tissues were observed with a fluorescence microscope BX53 (Olympus, Tokyo, Japan).

Statistical differences was analyzed by ANOVA with the Dunnett's multiple comparison test, or by Student's t-test. Correlation between the number of cells, and bioluminescence or fluorescence intensity was calculated by Pearson's product correlation coefficient.

We transfected the ZsGreen1-expressing vector into 5a-D-Luc cells derived from MDA-MB-231 (6,7), selected stably ZsGreen1-expressing cells, and then isolated five subclones. One of the five subclones that showed the strongest fluorescence intensity was selected for the following experiments and was named 5a-D-Luc-ZsGreen. In the 5a-D-Luc-ZsGreen cells, the signal of green fluorescence was observed in the cytoplasm (Fig. 1A) and the fluorescence intensity seemed to be stably maintained ≥12 passages in vitro (data not shown). There were no morphological differences between MDA-MB-231, 5a-D-Luc, and 5a-D-Luc-ZsGreen cells (Fig. 1A). The 5a-D-Luc and 5a-D-Luc-ZsGreen cells proliferated more rapidly than the parental MDA-MB-231 cells (P<0.01) (Fig. 1B). No significant difference in growth rate was observed between 5a-D-Luc and 5a-D-Luc-ZsGreen cells (Fig. 1B).

The bioluminescence and fluorescence intensities of 5a-D-Luc-ZsGreen were measured in vitro. Both bioluminescence and fluorescence intensities were detected when >10,000 cells were used. The number of cells was significantly correlated with bioluminescence (R² = 0.92, P<0.01) and fluorescence intensity (R² = 0.99, P<0.01) (Fig. 2A and B).

To evaluate the correlation of the number of cells with the bioluminescence and fluorescence intensities in vivo, 3 h after subcutaneous inoculation of 5a-D-Luc-ZsGreen cells into nude mice, the bioluminescence and fluorescence intensities were measured. Both bioluminescence and fluorescence intensities were detected when ≥10,000 cells were injected (Fig. 2C and E). There were significant correlations between the number of cells, and bioluminescence (R² = 0.89, P<0.01) or fluorescence intensity (R² = 0.84, P<0.01) (Fig. 2D and F).

To compare the metastatic ability between 5a-D-Luc and 5a-D-Luc-ZsGreen, these cells were injected into the left cardiac ventricle. As shown in Fig. 3A, bioluminescence signals were detected at 7 days after inoculation with 1×10⁵ cells, and the signal intensity and the number of metastatic sites increased over time. The signal intensity in metastatic regions in 5a-D-Luc-ZsGreen-inoculated mice became stronger than that in 5a-D-Luc-inoculated mice 21 and 28 days after inoculation (P<0.01) (Fig. 3B). According to ex vivo imaging and histological analysis, the 5a-D-Luc-ZsGreen cells metastasized to bone (all of five mice), brain (all of five mice), adrenal gland (two of five mice), and lymph node (one of five mice) by day 28 after intracardiac inoculation. The 5a-D-Luc cells metastasized to bone (all of five mice) and brain (two of five mice), other tissue metastasis was not observed by day 28 after intracardiac inoculation.

In all of the 5a-D-Luc-ZsGreen-injected mice, bone metastases in femora, tibiae, and spine were observed. Other bone metastases were also observed in mandibles (two of five mice), scapula (three of five mice), wrist joint (three of five mice), and incisor (two of five mice). The ZsGreen1 fluorescence was clearly detected in the metastasized regions by ex vivo imaging by fluorescence stereomicroscopy (Fig. 4A). The ZsGreen1 fluorescence was detected in the sections of metastasized bone after formalin fixation and decalcification (Fig. 4B and C). By the histological analysis of bone metastasis sites, bone marrow cavity was almost completely replaced by metastatic tumor cells at day 28 after intracardiac injection (Fig. 4B). Fluorescence imaging and TRAP staining showed clearly many osteoclasts were localized between the tumor cells and bone (Fig. 4C).

As shown in Fig. 5A, the green fluorescence of the tumor cells was widely distributed in the whole brain just after intracardiac injection. At 7 days after injection, only small and weak fluorescence foci were detected, and subsequently the intensity and size of metastatic foci increased over time (Fig. 5A). These findings were consistent with the results in bioluminescence imaging as shown in Fig. 3. The 5a-D-Luc-ZsGreen cells developed multifocal metastasis in the cerebrum, the cerebellum, and leptomeninges. This feature is common to brain metastasis in breast cancer patients (14).

Immunofluorescence staining of CD31 showed a close association of vessels with tumor cells. At 5 min after injection, tumor cells were detected within brain microvessels, especially at vascular branching points (Fig. 5B, upper left panel). At day 14, tumor cells were extravasated into surrounding tissues, and no cell was observed within the blood vessels (Fig. 5B, upper right panel). At day 21, tumor cells proliferated along microvessels (Fig. 5B, lower left panel), and alterations of vessel morphology and density were also observed at day 28 (Fig. 5B, lower right panel). By immunostaining for claudin-5, a tight junction marker used for evaluating the integrity of blood-brain barrier (BBB), the lack of claudin-5 expression was noted in a part of microvessels near tumor metastatic sites at day 28 (Fig. 5C).

Finally, in vivo two-photon laser-scanning microscopy allowed real-time tracking of circulating tumor cells in bloodstream of the brain. The tumor cells traveled through the bloodstream in brain vessels just after intracardiac injection (Fig. 5D). At 20 sec after injection, some tumor cells were plugged in a blood vessel branching point (Fig. 5D), and most of these cells disappeared by day 5 after injection (data not shown). Temporal imaging using two-photon microscopy revealed that: i) tumor cells proliferated in the brain parenchyma and microvessels formed in a single direction, and ii) microvessel density increased as tumor cells proliferated until day 21 after intracardiac injection (Fig. 5E).