All experiments were performed according to the "Principles for Research Involving Animals and Human Beings Guidelines" (OUHSC Animal Care & Use Committee protocol # 04–028). Double transgenic mice were obtained by crossing κB-lacZ with UPKII/SV40T mice. The κB-lacZ transgenic model used in this study was first described in 1996 17 and enriched in the C57Bl/6 background. It was constructed using the promoter of the gene encoding p105, a precursor of the p50 subunit of NF-κB. This promoter contains three NF-κB binding sites in its proximal part driving the expression of lacZ with a nuclear localization sequence. These mice permit the visualization of lymphatic endothelial cells in the urinary bladder and all other tissues examined so far 17.

A double transgenic mouse model (SV40-lacZ) was generated by crossing the κB-lacZ transgenics with UPKII/SV40T mice. UPKII/SV40T mice present the SV40 large T antigen under the control of a urothelium-specific mouse uroplakin II promoter, and develop specifically bladder cancer 121819. The phenotype of each mouse was confirmed by southern blot analysis of SV40T and PCR for β-galactosidase in tail vein snips. Only double positive mice were used in these experiments. FVB mice were purchased from Jackson Labs, crossed with κB-lacZ, and used as controls for UPKII/SV40T mice.

We follow the "Reporting Recommendations for tumor Marker (REMARK)" guidelines 20. This consensus report aims to lower the methodological variability of lymphangiogenesis quantification in tissue sections. The recommendations include: 1) double-blinded experiments; 2) the use of multiple IHC stains on serial sections (We used LYVE-1 IHC that has been established in our laboratories 11 and X-gal staining for β-galactosidase, because the SV40-lacZ enable this stain for visualization of lymphatic vessels 11); and 3) the use of a double immunostaining of lymphatic marker and those for cell proliferation. In this context, bladder cross-sections were double stained with LYVE-1 and Ki-67 antibodies.

LVD was determined as described before 1117. Eight mice per point were euthanized with sodium pentobarbital (100 mg/kg, i.p.) and tissues were removed rapidly and stained as whole mounts with X-gal. β-galactosidase activity was revealed by X-gal staining at 30°C for 4–6 hours. Whole mounts were examined under a dissecting scope (SMZ 1500, Nikon). All tissues were photographed by a digital camera (DXM1200; Nikon). Exposure times were held constant when acquiring images from different tissues. Afterwards, tissues were washed four times in PBS and post-fixed in 2% paraformaldehyde in PIPES [piperazine diethanesulfonic acid]. Lymphatic vessel density was quantified by morphometric analysis using Neurolucida workstation (MicroBrightField, Inc) 21, as described (ref 11; supplemental material). Morphometric analysis was performed by importing Neurolucida tracings into NeuroExplorer software version 3.70.2 (MicroBrightField, Inc) 21. The ratio between the area occupied by lymphatics and the total tissue area (μm²) was then determined for each section and statistical analysis was performed using a Wilcoxon's rank sum test. Results are expressed as mean ± SEM. In all cases, a value of p < 0.05 was considered indicative of a significant difference 22. This method may result in the overestimation of the lymphatic area. The software uses the 3D surface area of lymphatics, estimated from a 2D projection. Also, tissue area was calculated from a 2D projected cross-sectional tissue area (tissue contour).

Frozen bladders were processed for routine IHC according to published methods 23. Frozen sections were post-fixed in 1% formaldehyde. All reagent incubations and washes were performed at room temperature. 5% normal donkey blocking serum (Jackson Immunolabs) was placed on all slides for 45 min and sections were incubated with primary antibody for 1 hour and 45 minutes in a humidifier chamber. Slides were washed 3× 5 minutes in PBS and incubated with secondary antibodies. Slides were washed, counterstained with DAPI [4',6-diamidino-2-phenylindole; 1:20,000 dilutions of 10 mg/ml] for 2 minutes and coverslipped with Shur/Mount (TBS) mounting media and sealed with nail polish. All tissue cross-sections were visualized with a Nikon Eclipse TE 2000-S inverted fluorescent microscope (Nikon) 24 and imaged at room temperature using a digital CCD camera (Roper Scientific) 25, driven by NIS-Elements AR2.3 Imaging software (Laboratory Imaging/Nikon) 26. Controls included omission of the primary antibody. Antibodies used in this work are listed in Table 1. All secondary antibodies were used at a 1:400 dilution. Secondary antibodies included donkey anti-rabbit IgG AF488 conjugate (Molecular Probes; probes.invitrogen.com), donkey anti-goat IgG (Alexa Fluor 546, A11056, Invitrogen), and donkey anti-rat IgG (AlexaFluore 488).

At least 6 random fields per cross-section were visualized at 20× magnification and used for image analysis that was performed with the NIS-Elements Advanced Research 2.3 imaging software 26. This software identifies signal by thresholding key intensity values. Further the software permits imposing restrictions to the measurements by excluding false positive signals. Briefly, the number of positive cells expressing a particular antibody was calculated as percent of the region of interest (ROI), as indicated in the individual figure legend. Co-localization of two antibodies was calculated by converting the area occupied by cells positive for the first antibody into a ROI. Then the percent of cells that were positive for the second antibody was calculated within the ROI. Results are expressed as mean ± SEM. In all cases, a value of p < 0.05 was considered indicative of a significant difference 22.

Mice were fed a low-chlorophyll diet for 2 weeks to reduce auto-fluorescence in the intestinal region 27 and the abdominal hair was removed. Mice were anesthetized with isoflurane, placed in a heating pad, and received 200 μL of the Gd-Cy5.5 intravenously, and the accumulation of Cy5.5 into the urinary bladder was followed over time. Anesthetized mice were immediately placed on a heating pad inside a FluorChem HD2 (Alpha Innotech, San Leandro, CA) equipped with a Chromalight^® multi-wavelength illuminator and a 4-million pixel Cooled camera (F2.8 manual zoon lens and F1.4 fixed lens) coupled to a dedicated computer. The FluorChem cabinet permits continued anesthesia with isoflurane. Images were first acquired and stored with AlphaEase FC^® 32-bit software (Alpha Innotech, San Leandro, CA) and subsequently, application of black-and-white and color gradients were performed in Adobe Photoshop^® CS3 extended 28 that permitted the determination of integrated density. For this purpose, an elliptical marquee of fixed size (180 × 180 px) was used to determine the region of interest (ROI) around the luminescence zones corresponding to a bladder area and the count tool was used to determine and record each integrated density. The integrated density corresponded to the sum of the values of the pixels in the ROI, which were equivalent to the product of the area (in pixels) and mean gray value.

All data was acquired using a Bruker 7 Tesla/30 cm horizontal bore magnet. Anatomical scans acquired proved to be useful in elucidating normal and bladder tumor tissue. Bladder and tumor physiology was followed in a time course study using this method at each time point, for each animal. This method is a dual spin-echo technique modified from a MSME (Multi-Slice Multi-Echo) technique designed to yield anatomical T1 and T2 weighted images for each slice position in the slice package. Axial scans acquired had the following parameters: TR = 900 ms, TE = 11.6 ms, slice thickness = 0.75 mm, NA = 4, slice gap = 0.05 mm, FOV = 2.5 cm × 2.5 cm, matrix size = 256 × 256, giving an in-plane resolution of 98 μm × 98 μm. T1-weighted images had an effective TE of 17 ms, and T2-weighted images had an effective TE of 52 ms. Acquired parameter images had TE = 15 ms over a range of 5 TR increments with values = (100, 300, 450, 600, 850, and 1150 ms), slice thickness = 0.75 mm (position of slice varied according to tumor location), FOV = 2.5 cm × 2.5 cm, NA = 2, matrix size = 128 × 128, and an in-plane resolution of 195 μm × 195 μm. Mono-exponential fits were utilized to calculate actual T1 values in post-processing.

The macromolecular contrast material, biotin-BSA-GdDTPA, was prepared by the modification of the method of Dafni and collaborators 29. Bovine serum albumin (BSA, 0.5 g, 8 μmol; Sigma) was dissolved in 0.1 M sodium bicarbonate (7.5 ml, pH 8.5). Sulfo-NHS-Biotin (22.4 mg; 53 μmol; Pierce) was dissolved in double distilled water (DDW, 1.2 ml) and was added to BSA while stirring. The reaction mixture was stirred for 1 hr at 4°C and an additional 2 hrs at room temperature. The dialyzed product in 0.1 M Hepes buffer (pH 8.8) was reacted with diethylene triamine pentaacetic acid anhydride (DTPA, 0.5 g, 1.4 mmol; Sigma) suspended in 2.5 ml of dimethyl sulfoxide (DMSO) at room temperature. DTPA was added in portions and the pH was adjusted immediately after each addition to 8.5 with 5 N NaOH. The reaction mixture was stirred for 2 hrs at 4°C and extensively dialyzed against cold 0.1 M citrate buffer (pH 6.5). Finally, gadolinium (III) chloride (GdCl₃, 0.25 g; 0.67 mmol; Sigma) in 2.5 ml 0.1 M sodium acetate buffer (pH 6.0) was added gradually, and the mixture was stirred for 24 hrs at 4°C. The product, biotin-BSA-GdDTPA, was extensively dialyzed against cold citrate buffer (0.1 M, pH 6.5) and then against DDW. The product was lyophilized and stored at 4°C. For Cy5.5 -Gd preparation, 10 mg of dry lyophilized product (biotin-BSA-GdDTPA) was reconstituted in 150 μl of sodium bicarbonate buffer 0.1 M pH 8.8 NaHCO₃. Cy5.5 dye (Cy5.5 mono functional reactive dye, Amersham Biosciences) was dissolved in 10–20 μl of DMF and added to the product. The mixture was incubated in the dark at RT for 1 hr while mixing. The final compound biotin-BSA-GdDTPA-Cy5.5 was purified using Zeba 2 ml columns and had a molecular weight of ~82 kDa.

Dextran (500,000 MW)-Cy5.5 (Dex 0003–5) was purchased from Nanocs 30.

As the parent mouse model (κB-lacZ) permitted the visualization of lymphatic vessels by the expression of β-galactosidase, we investigated whether the double transgenic maintained this capacity. Figure 1A is a representative whole mount X-gal staining of bladders isolated from SV40-lacZ indicating that these mice maintained the constitutive activity of NF-κB in lymphatic vessels. Positive X-gal staining was also observed in cross sections left unstained (Figure 1B) or counter stained with H&E (Figure 1C and 1D). High magnification pictures of bladder cross sections indicate a nuclear localization of β-galactosidase expression in lymphatic endothelial cells due to a nuclear localization sequence of the original construct of the transgenic mouse (Figure 1D). Double staining of whole mounts confirmed that all β-galactosidase positive vessels were also LYVE-1-positive, and therefore, were lymphatic vessels (Figure 1E). In addition to lymphatic endothelial cells, cells morphologically resembling macrophages were the only other cell type that also showed LYVE-1 staining (Figures 2 A–C). This was further supported by the fact that these macrophage-like cells expressed MAC3, a macrophage marker 31 (Figures 2D–F). In contrast, other inflammatory cells were negative for LYVE-1 IHC. This is the case of mast cells present in the mouse bladder (Figure 2G–H).

Lymphatic vessel density was determined in control (FVB-κB-lacZ) and SV40-lacZ mice by Neurolucida^® guided morphological analysis of LYVE-1 IHC stained bladder whole mounts, as described (supplemental material to manuscript 11) and in cross-sections by fluorescent IHC. Figure 3A and 3B are representative LYVE-1 IHC of a urinary bladder isolated from SV40-lacZ mice. LYVE-1 staining was correlated with β-gal expression leading to results similar to those presented in Figure 1E and, therefore, indicating that the vessels represented in Figure 3A are indeed lymphatics. After IHC with LYVE-1 (Figure 3A) Neurolucida^® was used to trace and quantify the region of interest (ROI = white dotted line) and volume of lymphatic vessels (Figure 3B). Figure 3C presents the lymphatic vessel density (percent per unit of area) indicating that compared to 6-months old FVB-κB-lacZ mouse (n = 12), SV40-lacZ mice (n = 4 for each time point) presented a significant increase in LVD in the adventitia during cancer progression (BC = mean LVD 39.3% (n = 20); and control (FVB-κB-lacZ mice) = mean LVD 24.3%, [n = 12]; p < 0.001). This increase in LVD was observed in as young as 4-month old mice and remained elevated in all age groups studied so far (up to 9 months). Because LVD in whole mounts could only take into consideration medium and large lymphatics, an additional group of mice had their bladders removed and their cross-sections were studied by fluorescent IHC using LYVE-1 antibody. Image analysis of the cross-sections of 9-month old FVB-κB-lacZ and SV40-lacZ mice (ages 7, 9, and 11 months) are presented in Figure 3D (a total of 11 control and 12 SV40-lacZ cross-sections were examined), confirming a significant increase in LVD during bladder cancer progression.

Next, the same cross- sections used for quantification of LYVE-1 were double labeled with LYVE-1 and an anti-goat polyclonal antibody against Ki-67(M-19), a nuclear protein expressed in proliferating cells that may be required for maintaining cell proliferation 3233, and that provides the same value for proliferation index as BrdU in urothelial cancer cells 33. Figure 4A–D is a representative photomicrograph of sub-urothelial lymphatic vessels illustrating the finding that some of the lymphatic endothelial cells were positive for both LYVE-1 and Ki-67. These results indicate the presence of cancer-induced lymphangiogenesis in this mouse model.

For NIRF, a total of 9 SV40-lacZ mice ages 6–11 months were used. Mice were fed a low-chlorophyll diet for 2 weeks to reduce auto-fluorescence in the intestinal region 27 and the abdominal hair was removed. Mice were anesthetized with isofluorane and received 200 μL Gd-Cy5.5 intravenously (dose of 500 mg/kg), and the accumulation of Cy5.5 into the urinary bladder was followed over time.

A time-course of Cy5.5 accumulation in the lower abdominal region of SV40-lacZ is illustrated in Figures 5A–5F. Additional mice, represented in Figures 5G–J, have their abdomen opened and the gastrointestinal tract removed to permit better visualization of the pelvic floor. Near infra red fluorescence of the regions within red circles on Figures 5A–F is represented as integrated density in Figure 5K. The numbers in each segment indicate the time lapse after i.v. administration of Gd-Cy5.5. An absence of specific fluorescence in the lower abdominal regions of mice at zero and 10 min after Gd-Cy5.5 administration (Figure 5A and 5B) is observed. A gradual accumulation of Cy5.5 was observed and a fluorescence peak occurred in the bladder region at 120 minutes (Figure 5B and 5F). At 240 minutes a significant amount of fluorescence still remained in the urinary bladder (Figure 5C and 5G). At 330 minutes the urinary bladders revealed low NIRF (Figure 5E and 5I) when compared to the fluorescence peak observed at 120 minutes, and at 24 hours (1440 min) only residual fluorescence was observed (Figure 5J). In addition, it should be noted that emptying the residual urinary content of the bladder did not alter significantly the intensity of fluorescence, indicating that the residual fluorescence corresponded to Cy5.5 which still remained within the bladder parenchyma. Outside the urinary bladder, Cy5.5 accumulated in the kidneys (green arrows) and in lymph sacs (blue arrows in Figures 5G, 5I and 5J).

To confirm the results obtained in vivo, groups of mice (n = 3) were euthanized at 2, 5.5, and 24 hours (120, 330, and 1440 minutes) after i.v. administration of Gd-Cy5.5 and the bladders were removed for analysis of co-localization of Cy5.5 with LYVE-1- and CD31-positive cells. Figures 6, 7, 8 contain representative photomicrographics of confocal analysis of bladder cross-sections and the results of image analysis performed in all sections are presented in Figure 9. In all figures, the red represents Cy5.5 accumulation and the blue stain is DAPI (4',6-diamidino-2-phenylindole), which highlights the cell nuclei. The yellow label indicates co-localization of Cy5.5 with CD31-positive blood vessels, or LYVE-1-positive lymphatic vessels.

In particular Figures 6 A–B and 7 A–B indicate the sub-urothelial localization of blood vessels whereas lymphatic vessels are seen in the deeper regions of the mucosa but not in the subepithelial layer. Figures 8 A–D indicate that, at 1140 minutes, a significant amount of Cy5.5 still remains in the bladder parenchyma which may explain the fluorescence observed in Figure 5K.

Image analysis indicates that at all time points the area of the bladder cross-sections occupied by CD31-positive blood vessels was greater than the area occupied by LYVE-1-positive lymphatic vessels (Figure 9). In terms of co-localization of Cy5.5, at 120 minutes, Cy5.5 was found evenly distributed between LYVE-1-positive lymphatics and CD31-positive blood vessels (Figure 9). At 330 minutes a great proportion of Cy5.5 was found within LYVE-positive lymphatic vessels, and at 1140 minutes, most of the Cy5.5 was found within lymphatic vessels (Figures 9). These results provided the basis for subsequent MRI analysis of lymphatic vessel function.

A constant feature observed following i.v. administration of Cy5.5-conjugates was the labeling of abdominal lymphatic sacs (blue arrows in Figures 5G, 5I, and 5J). In order to investigate whether NIRF would permit visualization of these collecting lymphatic vessels and lymph sacs, high molecular weight dextran (500,000 MW)-Cy5.5 was injected in the pelvic floor underneath the urinary bladder, in the same region of lymph sacs of anesthetized mice and serial images were immediately captured. Figures 10 A–D are negative NIRF images acquired along with time and indicate the movement of Cy5.5 in major collecting lymphatic vessels and Figures 10E and 10F are higher magnifications of the same system. It was noted that a series of lymph sacs were observed squirting the fluorescent content rhythmically and upwards (black and white arrows indicate the direction of the flow in two major collecting lymphatic vessels). A quick time movie (Additional File 1), illustrates the upward movement of Cy5.5 within collecting lymphatic vessels shown in Figure 11.

Two hours after dextran-Cy5.5, the abdomen was opened and the gastrointestinal tract was removed, and representative photomicrographs were taken in black-and-white (Figure 12A) or artificially-colored to provide a gradation of fluorescence distribution (Figure 12B). It can be noticed that Cy5.5 accumulation occurred primarily in the bladder, uterus, and lymph sac or cysterna. NIRF also permitted the guided capture of tissues or organs with high fluorescence intensity. In this way, lymph sacs were dissected under fluorescence and processed for histology. Figure 12C is a composite of a cross-section of a lymph sac showing an interesting pattern of concentric paths of smooth muscle (black dotted circle) that can be appreciated at higher magnification in Figure 12D. A region with high concentration of lymphocytes was also visible (wavy black line on Figure 12C) and at the confocal level, it was possible to determine the presence of CD-31-positive blood vessels (Figure 12E; green) and LYVE-1-positive lymphatic vessels (Figure 12F; green) draining this region. In both figures, the red represents Cy5.5 accumulation and the blue stain is DAPI.

Although this work is very preliminary, it permitted our laboratory to identify and measure lymphatic function in vivo and in real time. These results indicate that NIRF can be use to study how Cy5.5-conjugate compounds are distributed between the interstitial fluid, blood vessels, and lymphatics. Figure 5K illustrates the kinetics of Gd-Cy5.5 and permitted the follow up of lymphatic vessel function by MRI (see below). In addition, large collecting lymphatics can be identified, the propulsion of lymph within lymphatic vessels can be visualized, and movies can be used to study these dynamics (Figure 11). Finally, NIRF-guided capture permitted us to start observing collecting lymphatics and lymph sacs and, therefore, to further study their structures.

Thanks to a unique rodent-dedicated MRI facility on campus 34, we were able to meet two major goals regarding the SV40-lacZ mice. The first goal was to detect the presence of BC as early as possible. This permitted a longitudinal study of tumor sizes during cancer progression. The second goal was to determine whether the observed increase in LVD was accompanied by an increased lymphatic function. A total of 7 SV40-lacZ mice and 3 wild type controls (FVB crossed with κB-lacZ mice) entered the study. Out of the 7 SV40-lacZ mice, 5 developed bladder tumors and 1 had to be euthanized.

We were successful in early bladder cancer detection, and MRI technology permitted the detection of bladder cancer as early as in 6 months (youngest animal scanned). Figures 13, 14, and 15 are representative double echo MRIs of 6-month old SV40-lacZ mouse bladder. Additional file 2 is a movie obtained during pre-contrast MRI indicating the different scanning positions (Figure 13). Additional file 3 is a post-contrast (Gd-Cy5.5) MRI movie that is highlighted in Figure 14. Figure 15 is a detail of the last frame of movie presented in additional file 3 (a red circle delimits the urinary bladder and waved green line delimits tumor areas that are darker than the normal tissue). Overall, the urinary bladders averaged volumes of 219.2 ± 74 mm³. In contrast, the total area of the tumors averaged 0.31 ± 0.1 cm², occupying areas of 31 ± 12.7 mm² and volumes of 15.6 ± 6 mm³. The smallest bladder tumor detected by MRI had a volume of 2.1 mm³. The tumor fractional volume in comparison with the whole bladder averaged 0.11 ± 0.02%. Longitudinal studies indicated that in some mice the tumor tripled its initial volume over 2 months (Figure 16).

Dynamic contrast-enhanced MRI (DCE-MRI) images with the use of the contrast agent Gd-Cy5.5 were acquired subsequent to anatomical scans providing sensitive regions of interest (ROI) for calculating T1 values between normal and bladder tumor relaxation times. We found an increased signal (due to decreased T1 relaxation) that correlated with an increased uptake of the Gd-probe from the blood vessels into the urinary bladder extra vascular space. The bladder volume did not change significantly during the imaging period. However, there may have been some motion artifacts associated with respiratory motion that was not compensated with respiratory gating. Figure 17A is a gray scale representation of peak accumulation of the Gd-probe in the bladder wall, except within tiny regions (a red circle delimits the urinary bladder and green lines indicated two tumor areas that are darker than the normal tissue). Figure 17B represents the difference image obtained at 30 minutes post-Gd-probe contrast subtracted from the pre-contrast image. This image illustrates an early enhancement by the contrast agent into the urinary bladder extravascular space, which may be attributed to vascular leakage. Figure 17C represents difference image obtained at 80 minutes post-contrast subtracted from 30 minutes post-contrast. This image demonstrates a delayed enhancement due to drainage and pooling of the contrast agent, which may reflect uptake within the lymphatics in the bladder. Figure 17D represents the difference image obtained between images in 17B and 17C (80 min. post-contrast minus 30 min post-contrast, and the 30 min. post-contrast minus the pre-contrast images). This image suggests that there is decreased drainage within tumorigenic areas and perhaps a pronounced increase in uptake in the lymphatics within normal regions.

Because drainage of macromolecules by the lymphatics is a slow event with rates of 2.3 to 3.7 μl/g minutes, images were acquired in two "phases" corresponding to the biphasic kinetics of the BSA-Gd-DTPA 15. The first or "early phase" is comprised of images obtained just before i.v. administration of 200 μL Gd-Cy5.5 (dose of 500 mg/kg). Starting at 3 minutes post-injection, images were acquired every 7 minutes, up to 2 hours post-contrast. The second block of MRI data was acquired up to 80 minutes after the first series of images. A digital subtraction of the images between pre- and 80 minutes post-contrast following Gd-Cy5.5 in three contiguous slices from the same bladder (positions -0.2, -0.8, and +0.8) was artificially colored (Figures 18A, 18B, and 18C). The red and yellow regions are active lymphatic vessels and consist of drainage events within the bladder interstitium. Figure 19 represents the comparison between tumor-bearing bladders (Figure 19A) and controls (Figure 19B and 19C). T1 values from these regions were normalized as percent of the reference site (muscle). This figure represents the time-dependent clearance of the contrast agent away from the bladder and provides a strong indication of LVF. Note the difference in T1 values between a tumor region and an adjacent normal bladder region (Figure 19A). In control bladders, two different normal regions were compared. These results suggested decreased drainage in the areas bearing tumors in comparison to normal areas.