A multipurpose reporter gene was introduced into the U87MG glioblastoma cell line by lentivirus transduction [22]. The hrLuc-RFP-TK tribrid gene [20, 27], courtesy of Sam Gambhir, UCLA (now at Stanford University), is a triple fusion of human codon-optimized Renilla luciferase (hrLuc), DsRed2 red fluorescent protein (RFP) (BD Clontech) and mutant sr39tk herpes simplex virus type 1 thymidine kinase (TK) (Ray et al. [20] for the precursor Rluc-sr39tk hybrid gene fusion article). Note that Ray et al. [20] report on an improved tribrid vector that replaces the DsRed2 RFP with a monomeric DsRed derivative (mRFP1, Campbell et al. [28]) and R.Y. Tsien’s lab has made spectral and brightness improvements to mRFP1 in their fruity fluorescent protein series of papers [29, 30].

Rluc enables whole animal in vivo bioluminescence imaging with a Xenogen 3D IVIS^® imaging system for quantitation of tumor cell mass following native coelenterazine injection (collaboration with Mike Rosol and Maya Otto-Duessel, CHLA). In this and related studies, tumor size was also estimated by T1- or T2-weighted or contrast enhanced (tumor tissue edema) MicroMRI using either a custom mouse coil on a clinical MRI machine (GE Medical Systems 1.5 Tesla MRI) or on a 7 Tesla Bruker MicroMRI (collaboration with Rex Moats, Harvey Pollack and Maya Otto-Duessel, CHLA). MRI cross-section area correlates well with histology area [22, 31].

The DsRed2 red fluorescent protein (RFP) enables whole mount confocal microscopy and fluorescence stereomicroscope (Leica MZ FL III) of split brains, fluorescence microscopy of tissue sections by DsRed2 fluorescence or anti-DsRed immunofluorescence or immunohistochemistry with a compound microscope (Leica DM RXA), and flow cytometry (BD Biosciences FACSCalibur) of trypsinized brain. The tumor cells are labeled with the tribrid hrLuc-RFP-TK fusion triple reporter because we wanted to reserve the more sensitive firefly luciferase (Fluc) in vivo reporter for therapeutic stem cell tracking [22, 29, 32, 33].

The TK gene product uses a ganciclovir substrate, which the herpes simplex virus mutant sr39tk thymidine kinase converts to a toxic product, and as a PET reporter (MicroPET collaboration with Xiaoyuan (Shawn) Chen, USC PET Imaging Science Center).

Mice were orthotopically injected with 10⁵ U87 glioblastoma cells. Mice were anesthetized using ketamine (Ketajet 100 mg/kg) and xylazine (Xyla-ject 10 mg/kg) and 1 × 10⁵ U87 MG glioblastoma cells in 1 μL of serum free medium were inoculated in 20 min stereotactically in the defined location of the caudate/putamen (0.5 mm anterior to the bregma, 2.0 mm lateral to the midline) using a 10 μL Hamilton syringe (Reno, NA) advanced to a depth of 3.3 mm from the cortical surface. The slow rate of 100,000 cells in 1 μL in 20 min was necessary for the tumor cells to remain deep in the brain tissue. Rapid infusion, or rapid drawback of the injection needle, can result in the tumor cells becoming dispersed along the needle track and rapid growth along the track and along the surface of the mouse brain (S. Yamada, V. Khankaldyyan, W. Laug, R. Moats, I. Gonzalez-Gomez, unpublished MicroMRI and histology data). A full stereotaxis rig maximizes consistency of the xenograft model (M. Rosol, M. Jensen, pers. comm.).

We used either 1 mg/mL fluorescein (isothiocyanate) conjugated tomato (Lycopersicon esculentum) lectin (Vector Laboratories), or Alexa Fluor^®-streptavidin (Molecular Probes) mixed immediately before use with biotin-conjugated tomato lectin (Vector Laboratories). Tumor-bearing mice were anesthetized and injected by cardiac puncture using a 28 gauge needle with 200 μL of fluorescent tomato lectin in PBS. Two minutes later, the mice were perfused with PBS intracardially to flush out red blood cells. Best results are obtained after maximally flushing out the red blood cells. An alternative to PBS perfusion (not used in the current work) is freshly prepared 4 % paraformaldehyde in PBS. This has the advantage of making the tumor tissue stiffer for hemi-sectioning, but has the disadvantage of increasing tissue auto-fluorescence, greatly increasing red blood cell auto-fluorescence, and disruption of some antigenic epitopes used for later immunofluorescence or immunohistochemistry. An alternative to cardiac puncture is tail vein injection. For lectin staining of brain microvessels, we obtained more consistent labeling with cardiac puncture. We have on occasion harvested other organs (liver, spleen, kidney/adrenal gland, lungs) from mice in this study (A. Yanai and V. Khankaldyyan) or from neuroblastoma orthotopic implantation or tail vein injection metastasis models (Chantrain et al. [34]; Y. DeClerck, C. Chantrain, K. Bajou, L. Sarte, S. Jodele, pers. comm.) after cardiac or tail vein injections. Cardiac puncture has a clear advantage for brain microvessel labeling, due to the “straight shot” from the aorta to the carotid arteries and brain blood vessels. Tail vein injection works somewhat better for the other organs, for those mice that have good tail veins and available injection sites following repetitive injections of chemotherapies or bioluminescent substrates. The neuroblastoma group has also evaluated Texas Red^®-streptavidin tomato lectin for microvessel imaging with similar performance to those reported here (DeClerck et al. and McNamara, unpublished). Animal protocols were performed with CHLA animal care committee and biosafety committee approval. We anticipate future experiments that will obtain similar or superior data with fluorescent nanocrystals (aka quantum dots), i.e., QD655-streptavidin and/or QD705 streptavidin (Quantum Dot Corp., http://​www.​qdots.​com) or Evitag720-streptavidin (Evident Technologies, http://​www.​evidenttech.​com), as has been published for blood (Larson et al. [3]) and lymph node (Kim et al. [35]) imaging in live animals [35].

For confocal imaging, brains were excised and sectioned in half (horizontal section). Horizontal sectioning was chosen because the U87MG-tribrid gene transfected human glioblastoma tumor cells were orthotopically injected in the middle of the mouse brain [21, 22, 26, 31]. The hemi-section was made with a scalpel, the cut being from the olfactory bulb to cerebellum, approximately bisecting the orthotopic implanted tumor. This is the same plane as one of the axes of the MicroMRI 3D scans [31], and is also used for histological sections by fluorescence immunohistochemistry and standard immunohistochemistry with either light microscopy (see below) or 35 mm film/Pathscan microscope slide scanner and image analysis [21, 34]. The brain halves were placed in cold PBS on ice, transported to the confocal microscope, transferred to a uncoated #0 or #1.5 coverglass 35 mm glass bottom dish (P35G-0-14-C, or P35GC-1.5-14-C, Mattek Corp., http://​www.​glassbottomdishe​s.​com/​gbcustomerpricew​eb.​pdf). Following confocal imaging, the brain hemi-sections were put back on ice, transported back to the wet lab, fixed, sectioned and then individual sections were processed for H&E histology, immunofluorescence and immunohistochemistry. In collaboration with Dr. Christine Brown, Renate Starr and Professor Michael Jensen, we have imaged Hoechst 33342 dye nuclear counterstaining of hemi-sectioned mouse brains, at the Light Microscopy Core of City of Hope National Medical Center. Brain nuclei were imaged on a Zeiss LSM 510 NLO confocal/multiphoton microscope in PBS with Hoechst 33342 at 10 μg/mL for 15 min, transferred to an imaging dish with PBS, and imaged using 750 nm, ~80 MHz, ~100 femtosecond pulses with a Zeiss 10×/0.5 NA lens. On the same microscope we have performed live cell experiments with Hoechst 33342 at 0.1 μg/mL in bicarbonate free, phenol red free, tissue culture medium. This concentration was chosen because higher concentrations of Hoechst are known to inhibit normal cell behavior [37]. When using Hoechst dyes, it is important to dilute the 10 mg/mL stock solution 100-fold in water because diluting in PBS results in formation of dye precipitates.

Details of the CHLA confocal microscope hardware can be found in the Appendix. The Leica SP1 confocal spectrophotometry hardware has been described [38, 39]. This and other spectral confocal microscopes is also available [40], where the CHLA system is system L4.

Most Images were acquired with Leica TCS SP1 confocal optics, equipped with air-cooled Argon ion (457, 476, 488, and 514 nm laser lines, ~2 mW power at 488 nm at the specimen plane), air-cooled Krypton ion laser (568 nm, ~2 mW power) and HeNe laser (633 nm, ~2 mW) mounted on a Leica DM IRBE inverted fluorescence microscope. The Argon and Krypton ion lasers operational lifetime is low if run at full power (maximum power knob setting of 4 o’clock on laser front panel); we routinely operated both at 12 o’clock power settings (mW power shown above for 100 % AOTF settings); when idle, the laser power was set to the minimum (knob setting 8 o’clock). Laser power was attenuated with the Leica SP1 AOTF. The Leica confocal microscope is maintained under an annual service contract with the manufacturer (Leica Microsystems, Exton, PA). In addition to an annual preventive maintenance visit by the manufacturer’s field service engineer, the image core manager performs periodic performance tests and arranges service visits as needed (testing details available from GM). Confocal system performance data, during the time period of image acquisition for this study, was published as system L4 [40].

Three reflection/fluorescence and one transmitted light (through the condenser) photomultiplier tubes (PMTs) are present on our SP1 confocal microscope. For fluorescence, the PMT offset were adjusted such that a positive intensity value was read out even with no light reaching the detector. Typical PMT gains were in the 700–900 settings range (maximum 1,250). A Leica RSP500 or TD488/568/633 triple dichroic mirror was used in the scanhead to reflect laser excitation to the microscope and pass emission photons to the PMTs. The SP1 optical head uses a prism spectral dispersion element with wavelength selection slits in front of each PMT. The SP1 is controlled from Leica LCS 2.5 on a Pentium 400 MHz computer. Confocal images were acquired with no filter cube in place. The Leica DM IRBE microscope stand is equipped Leica A, I3 and N2.1 longpass filter cubes, for DAPI, fluorescein/GFP/Alexa488, Cy3/DsRed, respectively, excited with a 50 W Hg lamp for visual inspection (a 100 W Hg voltage stabilized lamp would be a better choice for standard microscopy, since the Leica 50 W Hg lamp flickers). Information on filter cubes can be found in Ploem and Walter [41]. We occasionally acquired brightfield or Nomarski differential interference contrast (DIC) images with the laser line(s) from the objective lens and specimen being detected through the microscope condenser and a transmitted light path PMT that was positioned adjacent to the microscope transmitted light path, and switchable with a Leica knob. The transmitted light images were not confocal because that light path does not include a pinhole to reject out of focus light; the transmitted light images were in register with the confocal fluorescence. The laser illumination transmitted light DIC images do exhibit optical sectioning because of the intrinsic characteristics of DIC, but the images are not particularly good as the Leica optics exhibit significant shading, the contrast range of the 8-bit images are small, and the thickness of the hemi-sectioned mouse brains scatter much of the light.

Most brain images were acquired with a Leica 10×/0.40 NA HC Plan Apo Ph1 dry objective lens (Leica SP1 confocal microscope). Some images were acquired with a 10×/0.4 NA HC Plan Apo IMM lens, using water immersion, without significant improvement in image quality. We have tested a Leica N Plan 5×/0.12 NA, N Plan 2.5×/0.07 NA, and Plan Fluotar 1.3/0.04 NA objective lenses, for the unfixed hemi-sectioned brain application, but have been whelmed by the fluorescence brightness and contrast. These lenses may perform better using completely transparent, low-scattering, tissue with bright fluorescent markers, such as Zucker’s 1:1 benzyl benzoate–benzyl alcohol (BABB) clearing solution [8] or methyl salicylate clearing solution of brains perfused with red fluorescent latex vascular casts (S. Yamada, CHLA). BABB or methyl salicylate clearing of brain tissue requires ~1 week time after transfer to absolute ethanol, and leaves the tissue fragile with respect to later handling. By contrast, our unfixed brain hemi-section imaging adds a few hours delay between the time the mouse is sacrificed and when the tissue is ready for histological processing.

The confocal pinhole size was routinely set to either 1.0 or 1.5 Airy units when using the 10×/0.40 NA dry objective lens. At 1.0 Airy units, this lens has a theoretical XY resolution of 488 nm and Z resolution of 2,630 nm, with a working distance of 2,200 μm. (The deepest into a fluorescent specimen we have ever imaged was 836 μm with a methyl salicylate cleared red fluorescent latex mouse brain vascular cast.) Opening the pinhole from 1.0 to 1.5 Airy units results in the collection of more photons from the specimen from a thicker optical section. The image capture settings were 1,000 × 1,000 μm field of view (512 × 512 pixels), medium scan speed, 2 or 3 Kalman frame averaging. A Z step size of 2 μm (slightly oversampling in Z) was used to make the XY and Z pixel dimensions equal as a convenience for later image analysis and display (“orthogonal dimensions”). Our Leica SP1 confocal microscope is equipped with a Leica high speed, high resolution galvonometer based focus motor (“galvo-Z”). The galvo-Z enables high speed XZ scanning (at appropriate zoom) at the same speed as standard XY image scanning. The galvo-Z has a step size of ~0.040 μm, range of 160 μm and a mass limit of 90 g. For the purposes of this study, we used the Leica DMIRBE microscope internal focus motor, which has a nominal step size of ~0.100 μm.

Basic Leica SP1 microscope operations are described in a Web document by John Runions at the Haseloff lab (http://​www.​plantsci.​cam.​ac.​uk/​Haseloff/​JohnRunions/​Confocal_​instructions/​Confocal_​instructions(old)/​Confocal.​pdf). The Haseloff SP1 system is on an upright microscope and uses different lasers than the CHLA system.

Mouse skin blood vessels were imaged by sacrificing the mouse pinching the skin and cutting the skin scissors. For this work, an approximately 25 mm piece of skin (with white hair), was imaged in a 25 mm Mattek glass bottom dish (P35G-1.5-20-C, http://​www.​glassbottomdishe​s.​com) with a thin layer of PBS to improve optical contact to the coverglass. Tile scans were acquired with a Leica SP5/DMI6000 inverted confocal microscope (University of Miami). DiI perfused mouse skin blood vessels were acquired with a 561 nm laser, 10×/0.4 NA objective lens, 512 × 512 pixel images with 8 × 6 tile scan Z-series of 21 planes using 10 μm step size (200 μm thick). Non-confocal RGB transmitted light tile scans were acquired on the Leica SP5 microscope using 458, 561 and 633 nm laser lines for blue, green and red channels, respectively, using three scan track sequential frame mode (3× longer time than a single scan track). Because hemoglobin is the major absorbing molecule, in the red blood cells of the mouse that did not perfuse with DiI, we could have used any of the laser lines between 458 and 561 nm for the blue and green channels by applying a cyan lookup table, plus the 633 nm laser line for the red channel. This would have reduced scan time by one third (e.g., 90 min to 60 min for a 21 plane Z-series, 8 × 6 tile scan). For anyone performing transmitted light laser scanning diaminobenzidine and hematoxylin (DAB&H) tissue section slides, we recommend either the 405 or 458 nm laser line for DAB (blue channel) and 633 nm laser line for hematoxylin (green and red channels). Transmitted light laser scanning of hematoxylin and eosin (H&E) slides can performed with transmitted light 633 nm for hematoxylin, and either 488, 496, or 514 nm for eosin, with confocal fluorescence (520–560 nm) and/or non-confocal transmitted light detection (green and blue channel).

After brains were removed, bisected through the tumor, and imaged on the confocal microscope, they were processed, half for paraffin embedding (Zinc Fixative followed by 10 % formalin fixation) and half for frozen sections (4 % paraformaldehyde followed by 16 h in 30 % sucrose). Tissue for frozen sections were snap frozen in OCT embedding medium and maintained at −80 °C until sectioned. Hematoxylin and eosin stained sections were examined to note extent of tumor growth.

Immunofluorescence for CD31 and αSMA was performed on Zinc fixed paraffin-embedded sections. Five micrometer sections were mounted on Superfrost/Plus glass slides. After baking at 60 °C overnight, the slides were deparaffinized in xylene and rehydrated. For antigen retrieval, sections were boiled 12 min in Antigen Unmasking Solution (Vector Laboratories, Burlingame, CA). The sections were stained using anti-PECAM-1 (M-20) (Santa Cruz Biotechnology, Santa Cruz, CA) 1:100 and/or anti-Smooth Muscle Actin Clone 1A4 (DakoCytomation) 1:100 and incubated overnight at 4 °C. The secondary antibodies: Cy3-conjugated AffiniPure Donkey Anti-Goat IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) (1:200) and Fluorescein (FITC)-conjugated AffiniPure Donkey Anti-Rabbit IgG (Jackson ImmunoResearch Laboratories, Inc., West Grove, PA) (1:200). Frozen sections were stained with purified rat anti-mouse CD31 (PECAM-1) monoclonal antibody (BD Pharmingen) 1:75 Biotinylated anti-rat IgG (VECTOR Laboratories, Burlingame, CA).

Tissue immunofluorescence images were acquired with a Leica Microsystems (Bannockburn, IL) DMRXA fluorescence microscope, Sutter Instrument Company (Novato, CA) LS 300 Xenon arc lamp with 1 m liquid light guide, Life Imaging Services (Reinach, Switzerland) LLG-DMRXA quartz lens coupler, Chroma Technology Corp. (Rockingham, VT) HQ filter sets for fluorescein (set 41001), Cy3 (41007a), and Cy5.5 (41022). The microscope was also equipped with a 61000v2 DAPI/Ffluorescein/Cy3 triple pass filter set for viewing specimens by eye, and was equipped with additional filter sets for other applications. The DMRXA was equipped with a Leica optovar with 1× (lensless), 1.25× and 1.6× magnifications. Images were acquired using 100 % light to an Applied Spectral Imaging Inc. (Carlsbad, CA) SKY™ SD-300/VDS-1300 spectral imager, 12-bit digital CCD camera, using EasyFISH software. The camera was usually binned 2 × 2 to improve the signal to noise ratio, and decrease image exposure time, of the fluorescence images. All images were saved in 16-bit/channel quantitative EasyFISH file format, and in standard 24-bit TIFF format. For microvessel density measurements, slides were scanned at low power (HC Plan 10×/0.4 lens) to identify areas of highest vascularity. Ten to 20 high-powered (HC Plan 20×/0.7 lens) fields were then selected randomly within these areas, and microvessel densities were calculated based on the number of CD31/biotinylated tomato lectin positive structures. Microvessel counting was performed by multiple blinded observers in conjunction with a pathologist. A pericyte-positive vessel was defined as a CD31-positive vessel surrounded by at least one cell staining positive for α SMA.

For whole-section tumor area quantitation, H&E or immunohistology slides were acquired with a Pathscan adapter (Meyer Instruments) on a Polaroid SprintScan 4000Plus 35 mm slide scanner [23, 42]. See ref. 21 for examples of Pathscanned U87MG orthotopic tumor implantation model mouse brain tissue sections.

Leica Microsystems generously granted us permission to include the LCSLite software on the book CD (also available for free download from ftp://​ftp.​llt.​de/​softlib/​LCSLite/​ and the newer Leica LAS AF Lite is available for free download at ftp://​ftp.​llt.​de/​softlib/​LAS_​AF_​Lite/​). LCSLite operates in Microsoft Windows. The oldest version of LCSLite, v2.0.0871, runs under Windows NT, 2000 and XP. The newer 2.50.1347a and 2.61.1537 only run on Windows NT and XP. Windows Administrator privileges are required for installation. For best results, install on a computer with a dual monitor display and identical monitors. After running LCSLite (or LCS if you are a Leica confocal customer), by default the two channels in the demonstration datasets will open as green and red. Right clicking on the image window will bring up a menu from which you can turn on the LUT control. Clicking on the LUT slider allows you to change to gray or other lookup table color schemes. We recommend grayscale. Positioning the cursor over a LUT slider (but not clicking) results in thumbs being displayed along the top and bottom of the LUT—these allow you to adjust the contrast of a LUT channel. We suggest pulling the upper thumb down to increase the brightness of the DsRed2 channel. The overlay button can be used to display the combined colors.

Leica confocal microscope images are saved in tagged image file format (TIFF) with Z-plane and channel information as part of the filename. For our demonstration datasets, channel 0 (ch00) is PMT1 (fluorescein) and channel 1 (ch01) is DsRed2. Individual TIFF files can be opened in Adobe Photoshop CS, NIH Image, ImageJ, MetaMorph and other image analysis software packages. For ImageJ, we recommend the MBF ImageJ bundle (http://​www.​macbiophotonics.​ca/​downloads.​htm). For MetaMorph 6.25, we typically copy each channel set to its own subfolder, then use the Image Browser to open each channel as a separate stack. Some images were processed with Adobe Photoshop CS or MetaMorph 6.25 for publications.

The DiI labeling method [13] can also be applied to visualize blood vessels in the lungs or other tissues with the following procedure. An animal is killed by CO₂ overdose or sodium phenobarbital overdose (120 mg/kg, i.p) and then placed on a perfusion stage, positioned on its back. The trachea is exposed and cannulated with a blunt needle of #21-G. Cardiac perfusion is carried out as described (Li et al. [13]), but starts from the right ventricle instead of the left one and drains by cutting the descending aorta instead of the right atrium. The DiI solution is 3 mL with 30 μl of DiI stock solution. After perfusion, lungs are fixed with 4 % paraformaldehyde through the #21-G needle by connecting it to a small reservoir through a line under 20 mm H₂O intrapulmonary pressure (by raising the reservoir to 20 mm above the animal level) for 5–10 min. The lungs then are removed and submerged in 4 % paraformldehyde for 1 h. Tissue can be viewed directly under a fluorescence microscope to examine the surface vasculature. Thick vibratome sections (100 μm) are recommended to view vasculatures deeper than 100 μm from the tissue surface.

If perfusion fails with either tomato lectin or DiI, you can obtain useful information by transmitted light imaging, using hemoglobin absorption of the red blood cells, as your primary contrast. This is described in Subheading 3.5. The images will not be confocal optical sections, but you can still obtain useful information. We have successfully used RGB transmitted light mode Z-series tile scanning to acquire 200 μm thick Z-series from excised mouse skin.

When perfusing sacrificed mice, we encourage harvesting additional tissues and sharing them with colleagues interested in specific organs or tissues. You can contact your veterinarian about tissue sharing.