Noninvasive optical tracking of red fluorescent protein-expressing cancer cells in a model of metastatic breast cancer

Abstract

We have evaluated the use of the Xenogen IVIS 200 imaging system for real-time fluorescence protein-based optical imaging of metastatic progression in live animals. We found that green fluorescent protein-expressing cells (100 x 10(6)) were not detectable in a mouse cadaver phantom experiment. However, a 10-fold lower number of tdTomato-expressing cells were easily detected. Mammary fat pad xenografts of stable MDA-MB-231-tdTomato cells were generated for the imaging of metastatic progression. At 2 weeks postinjection, barely palpable tumor burdens were easily detected at the sites of injection. At 8 weeks, a small contralateral mammary fat pad metastasis was imaged and, by 13 weeks, metastases to lymph nodes were detectable. Metastases with nodular composition were detectable within the rib cage region at 15 weeks. 3-D image reconstructions indicated that the detection of fluorescence extended to approximately 1 cm below the surface. A combination of intense tdTomato fluorescence, imaging at > or = 620 nm (where autofluorescence is minimized), and the sensitivity of the Xenogen imager made this possible. This study demonstrates the utility of the noninvasive optical tracking of cancer cells during metastatic progression with endogenously expressed fluorescence protein reporters using detection wavelengths of > or = 620 nm.

Figure 1

Evaluation of the fluorescence intensity of three red fluorescent proteins and GFP when expressed in breast cancer cells with the Xenogen system. Shown on the upper panels of (A) are phase-contrast and corresponding fluorescence images of MCF-7 breast cancer cells transiently expressing mCherry, tdTomato, mPlum, and GFP or the empty vector. The numbers on the upper right corner of the fluorescent images are autoexposure times (in milliseconds) determined by the camera software. No autofluorescence was seen with either the GF cube or the RF cube with a 200-millisecond exposure of the MCF-7 cells transfected with the empty vector. The lower panels of (A) show Xenogen system images of the identical plate of cells shown on the upper panels of (A). False-color overlays represent the spectrum of photon radiance or flux (i.e., photons/sec/cm² of the imaged surface), radiating into a solid angle of 1 sr (Xenogen Living Image 2.5 manual). (The bar spectrum on the right of each image of the six-well plate is expressed in photons/sec/cm2/sr x 10⁶.) (B) Spectral scanning of red fluorescent proteins using the DsRed excitation filter and the 20-nm bandwidth emission filters centered on 600, 620, 640 660, 680, and 700 nm, as indicated on the lower right of each scan. (The bar spectrum on the right of each scan is expressed as photons/sec/cm²/sr x 10⁶.) (C) Phase-contrast and corresponding fluorescence image of MDA-MB-231 breast cancer cells that stably express tdTomato fluorescence protein and GFP. A uniform cell population with all cells fluorescing bright orange-red or green was observed. All fluorescence and phase-contrast photomicrographs were acquired at x 10 magnification on a Nikon ECLIPSE TS100 inverted microscope equipped with a CoolSNAP ES camera, and Texas red and FITC fluorescence filter cubes. Autoexposure times were determined by Image Pro Plus 5.1.

Figure 2

Evaluation of the detection of tdTomato and GFP fluorescence in the mouse cadaver phantom with the Xenogen system. The upper left panel of (A) shows fluorescence from tubes packed with 100 x 10⁶ MDA-MB-231-tdTomato-expressing (red) or MDA-MB-231-GFP-expressing (green) cells. Below these are false-color overlay images (exposure time, 0.01 second) acquired with the Xenogen system using either the GFP filter set or the DsRed filter set. The white tubes on these images indicate that tdTomato did not fluoresce when GFP was being imaged and that GFP did not fluoresce when tdTomato was being imaged. The mouse false-color overlay images (regions of interest were encircled; exposure time, 1 second) at the center and on the right of (A) show that the Xenogen system only detected tdTomato fluorescence from implanted tubes. (B) Similar results were found only in the number of tdTomato cells used (45 x 10⁶), whereas the number of GFP-expressing cells remained the same. Implanted tubes are shown on the left, where it is seen that red and green fluorescence were easily detected (exposure time, 0.01 second). (C) Autofluorescence from the fur of the SCID mouse depends on the emission wavelength filter used, as indicated on the bottom of each image. All three are unprocessed fluorescent images with a 2.5-second exposure time. Tube implants were those shown in (A) (left image) and (B) (center image), and in a tube containing 9.25 x 10⁶ tdTomato-expressing cells (right image). Regions of interest are encircled in the first two cases. The left and the center show the very intense fur autofluorescence that masks the detection of implanted fluorescence signals. The right image indicates that the use of a 620-nm emission filter allowed the lowest number of implanted tdTomota cells to be detected (fluorescent signal indicated by an arrow) above fur autofluorescence. (D) Excellent sensitivity of the Xenogen system. The false-color overlay on the left, with the region of interest encircled, shows the detection of the fluorescence signal of the 9.25 x 10⁶ tdTomato cell implant using a 0.01-second exposure and a 620-nm emission filter. (The bar spectrum below the image is expressed as photons/sec/cm²/sr x 10⁹.) The faint fluorescence from the implant has been made visible by enhancing the central image with PhotoShop, as indicated by the arrow on the image on the right.

Figure 3

Fluorescence optical tracking of the metastatic progression of MDA-MB-231-tdTomato in SCID mice. (A) Fluorescent images at 2 weeks postinjection of MDA-MD-231-tdTomato cells in either the first or the second mammary fat pad (bottom photographs). As depicted, the barely palpable invisible tumors (top photographs) were detected (center fluorescence images) with only a 0.25-second exposure time in the Xenogen system imager. Bright fluorescence spots on center images perfectly matched the sites of injection (bottom merged photographs). (B) Fluorescence and merged photographs of two representative mice at 3 weeks (top panel) and 8 weeks (bottom panel) post-MDA-MD-231-tdTomato cell injection. Caliper-measured tumor volumes are given at the bottom of the fluorescent images. Small tumors were easily imaged at 3 weeks and by 8 weeks, a metastatic lesion at a contralateral mammary fat pad had become visible with just a 1-second exposure time. The inset shows the image of the identical area after a 5-second exposure time. In addition, at 8 weeks, tumor sizes visualized with fluorescence imaging were in agreement with those measured with calipers; that is, the larger of the two tumors (157 mm³) is shown by fluorescence to be twice as large as the smaller one (82 mm³). (C) Fluorescent images of the same mice at 13 weeks post-MDA-MD-231-tdTomato cell injections with 1-, 2.5-, and 5-second exposure times. Nonfluorescent necrotic areas of the tumor are visible as hypointense patches in otherwise intensely fluorescent tumors. The arrows in the 1-second exposure images point to metastases to the axillary lymph node and contralateral mammary fat pad (dim images) of the mouse on the right and on the left, respectively. The primary tumor xenografts for the 2.5- and 5-second exposure times were covered with a blocking sheet to decrease bleedover fluorescence to nearby organs. Arrows in the 2.5-second exposure image clearly identify the contralateral mammary fat pad, as well as lymph nodes, in the other animal; arrowheads point out the faint detection of lung region metastasis. As shown, all metastases are easily identifiable as very bright areas of fluorescence on 5-second exposure images. (D) A fluorescent image after a 2.5-second exposure of the same two mice at 15 weeks after tumor cell injections. Tumor fluorescence shows that the tumor in the mouse to the left has regressed whereas the tumor on the mouse to the right has grown rapidly from Week 13. Caliper-measured volumes agree with fluorescence. Enlargement of the region of interest shows that the very large area of bright fluorescence in the pleural tissue/lung region is made up of clusters of smaller fluorescent masses (short arrows). Contralateral mammary fat pad metastasis is still apparent (long arrow). The central necrotic portions of the tumors appear as dark regions (arrowheads).

Figure 4

The incidence of lung metastases is independent of primary tumor growth rate. Left photograph: Gross morphology of lung metastasis following necropsy, which confirmed the nodular character of the lung and pleural tissue metastases observed during real-time imaging. Lung specimens from different mice are indicated by white numbers. Arrows show individual nodule locations in the respective lungs. The graph on the right shows tumor growth curves plotted as volume (mm³) against time postinoculation (weeks) for each mouse, as indicated by separate symbols. The images presented in Figure 2 were from two mice with the slowest-growing tumors, which indicates that primary tumor growth rate did not hinder metastatic progression.

Figure 5

Histologic confirmation of metastatic lesions formed by MDA-MD-231-tdTomato cells injected into the mammary fat pad of SCID mice. (A) Three representative examples of microscopic metastatic invasions to the lung. These lesions were identified using H&E staining (left photographs) and in consecutive sections with fluorescent microscopy (right fluorescent photomicrographs). As demonstrated by increasing magnification, the sensitivity of fluorescence imaging decreased with lower tumor burden. However, fluorescent microscopy facilitated the identification of low numbers of fluorescent tumor cells by eliminating the need to score for specific tumor markers. (B) H&E staining of a section of the contralateral mammary fat pad with metastatic invasion. The asterisk indicates tumor cells adjacent to ductal structures (arrow), as well as adipose tissues (arrowheads). (C and D) H&E staining of representative axillary lymph node sections from lymph nodes excised from above a tumor (C) or contralateral to the tumor (D). It can be seen that tumor cell invasion was extensive (asterisks) and has completely abolished lymph node structure.