intravital dual-colored visualization of colorectal liver metastasis in living mice using two photon...

Intravital Dual-Colored Visualization ofColorectal Liver Metastasis in Living Mice UsingTwo Photon Laser Scanning MicroscopyKOJI TANAKA,1* YUHKI MORIMOTO,1 YUJI TOIYAMA,1 YOSHINAGA OKUGAWA,1

YASUHIRO INOUE,1 KEIICHI UCHIDA,1 KAZUSHI KIMURA,2

AKIRA MIZOGUCHI,2 AND MASATO KUSUNOKI11Department of Gastrointestinal and Pediatric Surgery, Mie University Graduate School of Medicine,2-174 Edobashi, Tsu, Mie 514-8507, Japan2Department of Neural Regeneration and Cell Communication, Mie University Graduate School of Medicine,2-174 Edobashi, Tsu, Mie 514-8507, Japan

KEY WORDS colorectal cancer; liver metastasis; two photon laser scanning microscopy

ABSTRACT A major challenge of cancer biology is to visualize the dynamics of the metastaticprocess in secondary organs at high optical resolution in vivo real-time. Here, we presented intravi-tal, dual-colored imaging of liver metastasis formation from a single cancer cell to metastatic colo-nies in the living liver of living mice using two photon laser scanning microscopy (TPLSM). Red flu-orescent protein expressing murine (SL4) or human (HT29) colorectal cancer cell lines were inocu-lated to the spleen of green fluorescent protein expressing mice. Intravital TPLSM was performedby exteriorizing and fixing the liver lobe of living mice. This was repeated several times for thelong-term imaging of the same mouse. Viable cancer cells in the living liver of living mice werevisualized intravitally at a magnification of over 6003. Single cancer cells were arrested within he-patic sinusoids 2 h after injection. Platelet aggregation surrounding a cancer cell was observed,indicating a phenomenon of tumor-cell induced platelet aggregation. Cancer cells were extrava-sated from hepatic sinusoids to the space of Disse. Protrusions of Kupffer cells surrounding a can-cer cell were observed, indicating that Kupffer cells appear to phagocytose cancer cells. SL4 cellsformed liver metastatic colonies with extensive stromal reaction. Liver metastases by HT29 cellswere observed as a cluster of micrometastatic nodules. High-resolution, dual-colored, real-timevisualization of cancer metastasis using intravital TLPSM can help to understand spatiotemporaltumor-host interactions during metastatic processes in the living organs of living animals. Microsc.Res. Tech. 75:307–315, 2012. VVC 2011 Wiley Periodicals, Inc.

INTRODUCTION

Cancer metastasis is a complex, multi-step processthat includes cancer cell detachment from primarytumors, intravasation, entry into the circulation, extra-vasation, and the colonization of a distant organ(Nguyen et al., 2009). So far, the molecular mecha-nisms underlying this multi-step process have yet to beelucidated completely due to limitations of a molecularbiology based approach.

Intravital real-time imaging of cancer metastasis inthe living organ of living animals may enable us tohelp understand how cancer cells colonize in secondaryorgans or how they interact with host stromal cells(Kedrin et al., 2008). However, intravital imaging inliving animals at higher resolution and higher magnifi-cation is difficult due to poor sample stability, insuffi-cient tissue penetration, and autofluorescence of thetissue.

Two photon laser scanning microscopy (TPLSM) hasseveral advantages of high-resolution deep-tissueimaging up to near 1 mm, less phototoxicity and photo-bleaching, and simultaneously dual-colored imaging ofgreen fluorescent protein (GFP) and red fluorescentprotein (RFP) derivatives with different excitation andemission spectra by a single excitation wavelength,

compared with conventional confocal laser scann-ing microscopy (Quentmeier et al., 2009; Wang et al.,2010).

Recently, we have reported the challenging combina-tion of intravital TPLSM with an organ stabilizing sys-tem for in vivo real-time imaging of intraabdominalorgans (Toiyama et al., 2010). The fixation of the targetorgan used in our method minimized the microvibrationof the observational area caused by heart beat and respi-ratory movement, allowing intraabdominal organs in liv-ing mice to be visualized under intravital TPLSM.

Liver metastasis is the major cause of mortality inpatients with colorectal cancer. (Gallagher et al., 2010).To clarify the underlying mechanisms of this process, amorphological approach such as intravital imaging in

Additional Supporting Information may be found in the online version of thisarticle.*Correspondence to: Koji Tanaka, Department of Gastrointestinal and Pediat-

ric Surgery, Mie University Graduate School of Medicine, 2-174 Edobashi, Tsu,Mie 514-8507, Japan. E-mail: [email protected]

Received 14 January 2011; accepted in revised form 13 June 2011

Contract grant sponsor: Ministry of Education, Culture, Sports, Science andTechnology of Japan; Contract grant numbers: KAKENHI 22591484, 21591723and 21390377; Contract grant sponsor: Takeda Science Foundation

DOI 10.1002/jemt.21059

Published online 5 August 2011 in Wiley Online Library (wileyonlinelibrary.com).

VVC 2011 WILEY PERIODICALS, INC.

MICROSCOPY RESEARCH AND TECHNIQUE 75:307–315 (2012)

living animal model needed in combination with molec-ular biology.

Here, we present in vivo real-time images of tumor-host interactions in liver metastasis, and the develop-ment of metastatic formations from single cells tomicrometastatic colonies in living mice using TPLSMat high optical resolution and high magnification.

MATERIALS AND METHODSAnimals

Enhanced GFP-transgenic mice [TgN(b-act-EGF-P)Osb] of the C57BL/6 background were kindly pro-vided by Professor Masaru Okabe (Genome Informa-tion Research Center, Osaka University, Suita, Japan;Okabe et al., 1997). GFP nude mice (C57BL/6-BALB/c-nu/m-EGFP) were purchased from AntiCancer Japan(Osaka, Japan). Male GFP mice (20–22 g) were bred,housed in groups of six mice per cage and fed with apelleted basal diet (CE-7, CLEA Japan Inc., Tokyo, Ja-pan). Mice had free access to drinking water. Theywere kept in the animal house facilities at Mie Univer-sity School of Medicine under standard conditions ofhumidity (50 6 10%), temperature (23 6 28C) and light(12/12 h light/dark cycle) according to the InstitutionalAnimal Care Guidelines. The experimental protocolswere reviewed and approved by the Animal Care andUse Committee at the Mie University Graduate Schoolof Medicine.

Colorectal Cancer Cell Lines

RFP expressing murine (SL4) and human (HT29)colorectal cancer cell lines were purchased from Anti-Cancer Japan (Osaka, Japan). Cancer cells were grownin monolayer cultures in RPMI 1640 (Sigma-Aldrich,St. Louis, MO) supplemented with fetal bovine serum(FBS, 10% (vol/vol), GIBCO BRL, Tokyo, Japan), gluta-mine (2 mM), penicillin (100,000 units/L), streptomycin(100 mg/L), and gentamycin (40 mg/L) at 378C in a 5%CO2 environment. For routine passage, cultures werespilt 1:10 when they reached 90% confluence, generallyevery 3 days. Cells at the fifth to ninth passage wereused for liver metastasis experiments, which were per-formed with exponential growing cells.

Experimental Liver Metastasis

RFP expressing cancer cell lines were inoculated tothe spleen of GFP expressing mice. GFP mice wereused as a syngeneic tumor model for SL4. By contrast,GFP nude mice were used as a xenogeneic tumor modelfor HT29. Exponentially growing RFP-expressing can-cer cells were harvested with trypsin/EDTA, washed inserum-containing RPMI 1640 medium to inactivateany remaining trypsin. The cells were centrifuged andresuspended in PBS. Finally, the cells were adjusted to1 3 107 cells/mL for single cell suspensions. GFP micewere anesthetized with an intraperitoneal injection ofchloral hydrate (Sigma, St Louis, MO). Under directvision, 1 3 106 cells were injected into the spleen usinga 30-gauge needle through a small incision of left-lat-eral abdomen of anesthetized GFP mice.

Surgical Procedures for Intravital TPLSM

After inoculation, GFP mice were anesthetized withan intraperitoneal injection of chloral hydrate (Sigma,St Louis, MO). Body temperature was kept at 378C

throughout the experiments using a heating pad.Upper midline laparotomy was made as short as possi-ble (less than 15 mm). The left-lateral lobe of the liverwas identified and exteriorized through laparotomy.After exteriorization, the liver lobe was put on theorgan stabilizing system (patent number: 2007-129723) using a solder lug terminal with an instant ad-hesive agent. The organ stabilizer minimized themicrovibration of the observational area caused byheart beat and respiratory movement. The stabiliza-tion and fixation of the liver lobe was the most impor-tant and technical difficult part of the intravitalTPLSM setup. After the application of PBS to theobservational area, a thin cover glass was gently placedon top of the liver surface. After intravital TPLSM,exteriorized liver lobe was gently removed from theorgan stabilizing system using a release agent to pre-vent liver injury. Sodium hyaluronate and carboxyme-thylcellulose membrane (Seprafilm Adhesion Barrier,Genzyme Corporation, Cambridge, MA) was placedbetween the liver and abdominal wall to prevent post-operative dense adhesion.

Interval TPLSM

To observe the process of liver metastatic formationin the same liver of the same GFP mouse, intravitalTPLSM was repeated at several times (intervalTPLSM). In other word, Interval TPLSM consists ofseveral times intravital TPLSM on the same mouse.The above-mentioned surgical procedures of intravitalTPLSM were performed until nondissecting adhesionsformed between the liver and abdominal wall. Intravi-tal TPLSM with high-resolution images can be per-formed at several times over intervals ranging fromdays to months. Precautions were required during theentire surgical procedure to prevent postoperative in-traperitoneal infection.

Setup of TPLSM

Procedures for TPLSM setup were performed as pre-viously described (Toiyama et al., 2010). Experimentswere performed using an upright microscope (BX61WI;Olympus, Tokyo, Japan) and a FV1000-2P laser-scan-ning microscope system. The use of special stage risersenabled this unit to have an exceptionally wide work-ing distance. This permitted the stereotactically immo-bilized, anesthetized mouse to be placed on the micro-scope stage. The microscope was fitted with severallenses with high numeric aperture for the long workingdistances required for in vivo work and water-immer-sion optics. In TPLSM mode, the excitation source wasMai Tai Ti:sapphire lasers (Spectra Physics, MountainView, CA), turned and mode-locked at 910 nm. The MaiTai procedures light pulses of about 100 fs width (repe-tition rate 80 MHz). Laser light reached the samplethrough the microscope objective (603 LUMPlanFI/IR,water dipping, numerical aperture of 0.9, working dis-tance 2 mm), connected to an upright OlympusBX61WI microscope. Data were analyzed by FV10-ASW (Olympus, Tokyo, Japan).

Two-photon fluorescence signal was collected by in-ternal detector using the excitation wavelength of 910nm to enable the simultaneous acquisition of EGFPsignal and RFP (DsRed2) signal. Images, color-coded

Microscopy Research and Technique

308 K. TANAKA ET AL.

green, and red, respectively, were subsequently mergedinto a single image.

Imaging of Colorectal Liver MetastasisUsing Interval TPLSM

We firstly screened the surface of the liver lobe atlower magnification (3100) setting out the X/Y planeadjusting the Z axis manually to detect the optimal ob-servation area containing RFP expressing cancer cells(at least five areas). After such first screening at lowermagnification, we scanned each area of the interest athigher magnification (water immersion objective 603with or without 23 zoom) by the manual setting of theX/Y plane and adjustment of Z axis (either automati-cally or manually) to obtain the high-resolution andclear TPLSM images. The imaging depth or imagingstacks was determined arbitrarily for visualizing thephenomena of colorectal liver metastasis in vivo real-time, three-dimensionally. Laser power was adjusteddepending on the imaging depth.

Experimental Schedule of IntervalTPLSM With At least Three Times

of Intravital TPLSM

RFP expressing cancer cell lines were inoculated tothe spleen of GFP expressing mice. The inoculation ofRFP-SL4 cells in GFP mice was used as a syngeneic

tumor model. By contrast, the inoculation of RFP-HT29 cells in GFP nude mice was used as a xenogeneictumor model. Preliminary experiments indicated thatliver metastasis formation was observed by 2 weeks ina syngeneic tumor model, and also by 8 weeks in a xen-ogeneic tumor model (data not shown).

Therefore, the first-round intravital TPLSM was per-formed for imaging the early stages of colorectal livermetastasis at 2 h after inoculation. The second-roundintravital TPLSM was also performed for imaging theearly events of colorectal liver metastasis at 24 h afterinoculation on the same mouse. To image the estab-lished liver metastases, the third-round intravitalTPLSM was performed at 2 weeks after the inoculationof RFP-SL4 cells in GFP mice, or at 8 weeks after theinoculation of RFP-HT29 cells in GFP nude mice.

At the end of the experiments (at 2 weeks after inoc-ulation for a syngeneic model, and at 8 weeks afterinoculation for a xenogeneic model), the whole liverwas harvested and subjected to histopathologicalanalysis.

Immunohistochemistry of Cytokeratin 20

Because of the lack of immunological cross-reactivitywith other cytokeratins, CK 20 has become an impor-tant tool for delineating the origin of metastatic humanadenocarcinomas arising from an unknown primary

Fig. 1. Colorectal cancer cell lines under fluoromicroscopy in vitro. (A) RFP expressing HT29 cellsand (B) RFP expressing SL4 cells (bar, 50 lm).


309INTRAVITAL IMAGING OF COLORECTAL LIVER METASTASIS

source. Mouse liver was removed and fixed in 4% form-aldehyde in phosphate-buffered saline (pH 7.4) for 24h, processed, embedded in paraffin wax according tostandard procedure. Formalin-fixed, paraffin-embed-ded tissue was sliced at a thickness of 3 lm, and thesections were placed on silane-coated slides. Afterdeparaffinization and dehydration, the sections wereautoclaved for 10 min in 10 mM sodium citrate bufferfor antigen retrieval. They were blocked and incubatedwith primary antibody overnight at 48C. Primarymonoclonal antihuman cytokeratin 20 antibody (CloneKBsB20.8; DakoCytomation) was used at a dilution of1:50 for implementation of the labeled streptavidin-bio-tin method (LASB2 kit/HRP, DakoCytomation). Cyto-keratin 20 was detected by Envision reagents (Envi-sion kit/HRP, DakoCytomation, Denmark). The sec-tions were counterstained with hematoxylin. Negativecontrols were also run simultaneously with preimmuneimmunoglobulin.

RESULTSCancer Cell Lines Under

Fluoromicroscopy In Vitro

Figure 1 shows RFP expressing HT29 and SL4 underfluoromicroscopy in vitro. HT29 cells had an oval toround shape (Fig. 1A). By contrast, SL4 had a spindle-shaped morphology (Fig. 1B).

The Establishment of In Vivo Real-TimeImaging for Colorectal Liver

Metastasis by TPLSM

Figure 2 showed an overview of the liver lobe fixationand intravital TPLSM setup. Figure 3 also showed aschematic drawing of the liver lobe fixation by an organstabilizing system for intravital TPLSM.

There were several key steps to establish our experi-mental protocol. These were as follows: (1) the optimallaparotomy for liver imaging by intravital TPLSM(transverse incision or longitudinal incision), (2) thechoice of liver lobe for intravital TPLSM using anorgan stabilizing system (right lobe or left lobe), (3) thefixation of the liver lobe to an organ stabilizing system(the use of a solder lug terminal and an instant adhe-sive agent), (4) the adjustment of parameters for thesimultaneously dual-colored imaging of RFP express-ing cells in the liver of GFP mice (the adjustment oflaser power and detection sensitivity), (5) intervalTPLSM with at least three times of intravital TPLSM(the release of the liver lobe from a solder lug terminalusing a release agent, and the placement of SeprafilmAdhesion Barrier to the abdomen), and (6) the accom-plishment of all experimental steps for intravital andinterval TPLSM (the improvement of our technique tocomplete all experimental steps).

Each experimental step needed �10 mice to solve thetechnical problems and to obtain the acceptableresults. As a result, �80 mice were used for the estab-

Fig. 2. Overview of the liver lobe fixation and intravital TPLSM setup. (A) Upper midline laparotomyand the exteriorization of the left-lateral lobe. (B) The fixation of the left-lateral lobe to a solder lug ter-minal using an instant adhesive agent. (C) The placement of an anesthetized mouse on the stage and thesetup of an organ stabilizing system.



lishment of in vivo real-time imaging for colorectalliver metastasis by TPLSM. After completing thesesteps along with the acceptable intravital TPLSMimages, we started to study the intravital, dual-coloredimaging of the early and late stages of colorectal livermetastasis using TPLSM. Once we got the ability tocomplete intravital and interval TPLSM, our successrate of this experiment was �100% with much lesstechnical failure.

Normal Liver of Living MiceUnder Intravital TPLSM

Using an organ stabilizing system, the motion frombreathing or heartbeats had little effect on intravitalTPLSM image acquisition. As a consequence, high-re-solution images of normal liver were visualized in liv-ing GFP mice. TPLSM imaging can be represented as atime-lapse two-dimensional movie, and also a z-stacksthree-dimensional movie. The imaging depth wasdetermined arbitrarily for visualizing the metastaticphenomena in vivo real-time, three-dimensionally. Itdepended in part on the positioning of the liver lobeusing an organ stabilizing system or laser power.

Liver structures such as hepatocytes, hepatic sinu-soids, hepatic endothelial cells, and blood components,such as leukocytes and platelets within hepatic vessels,were clearly observed with acceptable motion artifacts(Fig. 4A and Supplementary movie 4A). Since red blood

cells were not identified in GFP mice (Okabe et al.,1997), leukocytes were recognized as larger round cells,and platelets were recognized as smaller ones withinhepatic vessels. Leukocytes including Kupffer cellswere rolling or flowing in hepatic sinusoids. Plateletaggregation was observed in some area of hepatic sinu-soids. This phenomenon may suggest the disorder ofhepatic blood flow due to the exteriorization of the liverlobe. Vitamin A simultaneously emitted green- andred-color at the excitation wavelength of 910 nm, caus-ing autofluorescence. Vitamin A was therefore identi-fied as yellow round dots in the merged image.

The Rate of Successful Intravital TPLSMat the Indicated Time Points

Table 1 showed the rate of successful intravitalTPLSM at the indicated time points in both syngeneicand xenogeneic tumor models. At 2 h after inoculation,the first-round intravital TPLSM was successfully per-formed for imaging the early stages of colorectal livermetastasis in all 20 GFP mice (100%). One of 20 micewas lost due to the unknown cause within 24 h afterinoculation in both models. The second-round intravi-tal TPLSM was also performed in the remaining 19GFP mice (100%) of both models.

By 2 weeks after the inoculation of RFP-SL4, 5 micedied of colorectal liver metastases before the scheduledthird-round intravital TPLSM. The remaining 14 GFP

Fig. 3. Schematic drawing of the liver lobe fixation by an organstabilizing system for intravital TPLSM. (A) A solder lug terminalwas adhered water-tightly to the surface of the left-lateral liver lobeusing an instant adhesive agent. (B) After applying PBS to the obser-vation area, a thin cover glass was placed water-tightly. (C) A solderlug terminal was hold on by an organ stabilizing system. The stabili-

zation and fixation of the liver lobe enable us to image the liver struc-ture in vivo real-time at higher optical resolution and higher magnifi-cation under TPLSM. (a) solder lug terminal, (b) instant adhesiveagent, (c) observation area, (d) cover glass, (e) PBS, (f) left-laterallobe, (g) objective, (h) water immersion, and (i) organ stabilizing sys-tem.



mice (100%) were imaged successfully by the third-round intravital TPLSM (100%).

By 8 weeks after the inoculation of RFP-HT29, 2mice died of colorectal liver metastases before thescheduled third-round intravital TPLSM. The remain-ing 17 GFP nude mice were imaged successfully by thethird-round intravital TPLSM (100%).

Viable Cancer Cells in theLiving Liver of Living Mice

Red-colored cancer cells were visualized in thegreen-colored liver structures of GFP mice in vivo real-time under intravital TPLSM. Single cancer cells were

arrested in hepatic sinusoids 2 h after injection andseen at a magnification of over 6003 (Fig. 4B and Sup-plementary movie 4B). During the observation time(�1 h per session), circulating or rolling cancer cells inhepatic vessels were never observed. Hepatic sinusoidsappeared to be occluded by cancer cells. These findingssuggest that the arrest of cancer cells may be caused bymolecular adhesion between cancer cells and hepaticendothelial cells or size-dependent occlusion.

Tumor Cell Induced Platelet Aggregation

Platelet aggregation surrounding a single cancer cellwas observed 2 h after injection (Fig. 4C and Supple-

Fig. 4. In vivo real-time imaging of colorectal liver metastasis inthe living liver of living mice. (A) Normal liver, (B) viable cancer cellsin hepatic sinusoids (2 h after inoculation), (C) tumor cell induced pla-telet aggregation (2 h after inoculation), (D) phagocytosis by a Kupffercell (24 h after inoculation), and (E) extravasation of a cancer cell (24h after inoculation; bar, 50 lm). We used a 360 water immersion lenswith a numerical aperture of 0.9. The working distance of this objec-tive was 2 mm long. The two-dimensional optical section was an X/Y

plane at 210 3 210 lm2 (water immersion objective 360) or at 105 3105 lm2 (water immersion objective 360 with 32 zoom). The locationof the focal plane (Z) was adjusted arbitrarily, depending on the posi-tioning of the liver lobe using an organ stabilizing system or laserpower. All supplementary movies are time-lapse (A) or z-stacks (B–E).The imaging stacks were as follows: (B) 40 lm z-stacks, (C) 50 lm z-stacks, (D) 30 lm z-stacks, and (E) 20 lm z-stacks. All the imagestacks were acquired at the surface of the tissue.

TABLE I. The rate of successful intravital TPLSM at the indicated time points

RFP-SL4 cellsin GFP mice

RFP-HT29 cellsin GFP nude mice

Intravital TPLSM at 2 h after inoculation 100%a (20/20)b 100% (20/20)Intravital TPLSM at 24 h after inoculation 100% (19/19) 100% (19/19)Intravital TPLSM at 2 weeks after inoculation 100% (14/14) Not availableIntravital TPLSM at 8 weeks after inoculation Not available 100% (17/17)

aPercentage of successful interval TPLSM at the indicated time points.bThe number of mice subject to intravial TPLSM/the number of mice available at the indicated time points.



mentary movie 4C). Platelet aggregation was fre-quently observed within hepatic sinusoids in normalliver, suggesting a disturbance to hepatic blood flowcaused by the exteriorization of the liver lobe. However,cup-like shaped platelet aggregation around a cancercell may indicate a phenomenon of tumor-cell inducedplatelet aggregation (Erpenbeck et al., 2010).

Phagocytosis by Kupffer Cells

Leukocytes, including Kupffer cells, were recognizedas larger round cells, and were rolling or flowing in he-patic sinusoids. Figure 4D showed that reticular pro-trusions from a leukocyte-like cell might be capturing acancer cell (Supplementary movie 4D). This leukocyte-like cell was thought to be a Kupffer cell which had anameboid morphology. Thus, this phenomenon may indi-cate that Kupffer cells phagocytose cancer cells in he-patic sinusoids.

Extravasation of Cancer Cells

The space of Disse is localized between hepatic endo-thelial cells and hepatocytes. Figure 4E showed that aspindle-shaped cancer cell was localized in this space(Supplementary movie 4E). RFP-SL4 cells are roundpostinjection, and change in morphology from a round-shape to a spindle-shape during metastatic coloniza-tion. Thus spindle-shaped RFP-SL4 cells were thoughtto be extravasated from hepatic sinusoids to the spaceof Disse 24 h after injection.

The Incidence of the Early Events of ColorectalLiver Metastasis on the Portal Route

Table 2 showed the incidence of the early events ofcolorectal liver metastasis on the portal route in bothsyngeneic and xenogeneic tumor models. To observethe early events of colorectal liver metastasis at higherresolution and dual-colored, we firstly detected at leastfive areas containing RFP expressing cancer cells atlower magnification (3100). Thereafter, we observedeach area of the interest at higher magnification (waterimmersion objective 603 with or without 23 zoom). Ifnecessary, additional five areas were observed by themanual setting of the X/Y plane and adjustment of Zaxis either automatically or manually.

The phenomena of tumor cell arrest/adhesion and tu-mor cell induced platelet aggregation was frequentlyobserved at 2 h after inoculation. The extravasation ofcancer cells and phagocytosis by Kupffer cells wasobserved at either 2 or 24 h after inoculation, but theywere extremely rare events.

Metastatic Colonization in the Liver

Liver metastases by RFP-SL4 cells were observedwithin 2 weeks following injection in GFP mice (a syn-geneic tumor model). By contrast, liver metastases byRFP-HT29 cells were observed within 8 weeks in GFPnude mice (a xenogeneic tumor model). Figure 5showed macroscopic (A), microscopic (B), and TPLSMimages (C) of liver metastases by RFP-SL4 and RFP-HT29 cells. RFP-SL4 cells formed liver metastaseswith extensive stromal reaction. By contrast, liver met-astatic nodules of RFP-HT29 cells were composed ofclusters of micrometastatic colonies (Fig. 3C).

Table 3 showed the rate of liver metastatic formationand its successful observation by intravital TPLSM.

At 2 weeks after the inoculation of RFP-SL4 cells, 14GFP mice were imaged successfully by the third-roundintravital TPLSM (100%; 14/14). Since RFP-SL4 cellsshowed the diffuse growth pattern, liver metastatic col-onies were also imaged by intravital TPLSM (100%; 14/14).

At 8 weeks after the inoculation of RFP-HT29 cells,17 GFP nude mice were imaged successfully by thethird-round intravital TPLSM (100%; 17/17, as shownin Table 1). Among these 17 GFP nude mice, 12 (70%)mice had liver metastases with localized growth pat-tern at 8 weeks after inoculation. The left-lateral lobeof the liver can be only imaged in our experimental pro-tocol. Four mice had liver metastases outside the obser-

TABLE II. The incidence of the early events of colorectalliver metastasis on the portal route

Early event

RFP-SL4cells in

GFP mice

RFP-HT29cells in GFPnude mice

Tumor cell arrest/adhesion 20/20a (100%)b 20/20 (100%)Tumor cell induced plateletaggregation

17/20 (85%) 15/20 (75%)

Extravasation from hepatic sinusoids 2/20 (10%) 1/20 (5%)Phagocytosis by Kupffer cells 2/20 (10%) 2/20 (10%)

aThe number of mice having the indicated phenomenon/the number of mice ino-culated the indicated cells.bPercentage.

Fig. 5. Colorectal liver metastases by each cancer cell line. (A)Macroscopic findings, (B) microscopic findings (bar, 50 lm), and (C)intravital TPLSM images (bar, 50 lm).



vation area. As a result, the only 8 (67%) out of 12 micewere successfully imaged by the third-round intravitalTPLSM (Table 3).

ImmunohistochemicalDetection of Solitary Cancer Cells

Antihuman cytokeratin 20 antibody was used for thedetection of RFP-HT29 cells in the xenogeneic liver me-tastasis model. Micrometastatic colonization and a soli-tary RFP-HT29 cell in hepatic sinusoids were shown inFigure 6.

DISCUSSION

This study showed high optical resolution, highermagnification (over 6003), dual-colored (red: cancercells, green: host cells) images of colorectal liver metas-tasis formation from a single cancer cell to metastaticcolonies in the living liver of living GFP mice underintravital TPLSM.

TPLSM has been used for in vivo imaging because ofits deeper tissue penetration with less phototoxicity(Quentmeier et al., 2009; Wang et al., 2010). In vivoimaging of these organs using TPLSM has, however,been an enduring technical challenge, as motion arti-facts due to cardiac and respiratory activity affecthigher resolution and higher magnification imaging ofintraabdominal organs including the liver (Starodubet al., 2008).

Previously, we reported a new method for in vivoreal-time imaging of intraabdominal organs underintravital TPLSM (Toiyama et al., 2010). In this study,we improved and refined our technique for exterioriza-tion of liver lobe through laparotomy and a fixation ofexteriorized liver with an organ stabilizing system.These technical refinements enabled in vivo real-timeimaging at higher optical resolution and higher magni-fication in the living liver of living mice with acceptablemotion artifacts. Furthermore, we can perform intravi-tal TPLSM at several times on the same liver lobe of

TABLE III. The rate of liver metastatic formation and its successful observation by intravital TPLSM

RFP-SL4 cells in GFP mice RFP-HT29 cells in GFP nude mice

Liver metastases after intrasplenic inoculation 100%a (14/14)b at 2 weeks 70% (12/17) at 8 weeksObservation of metastasis by intravital TPLSM 100%c (14/14)d at 2 weeks 67% (8/12) at 8 weeks

aPercentage of liver metastases after intrasplenic inoculation.bThe number of mice with liver metastases/the number of mice available at the indicated time points.cPercentage of successful observation of metastatic colonies by intravital TPLSM.dThe number of mice successfully observed metastatic colonies by intravital TPLSM/the number of mice with liver metastases.

Fig. 6. Immunohistochemistry of cytokeratin 20. Micrometastasis by RFP-HT29 cells (A: bar, 500lm; C: bar, 50 lm), Solitary RFP-HT29 cells in hepatic sinusoids (B: bar, 500 lm; D: bar, 50 lm).



the same mouse for long-term imaging of colorectalliver metastasis. This interval TPLSM allows the ob-servation of a single cancer cell to metastatic coloniesin the same liver of the same mice.

However, our method has several limitations. Firstly,motion artifacts could be minimized, but not com-pletely suppressed, which is critical for intravital imag-ing at a magnification of over 6003. Additional anes-thesia may improve the quality of the image at highermagnification. Secondly, it is possible that the exterio-rization of the liver may affect hepatic microcircula-tion, which in turn may affect the process of liver meta-static formation at the cellular level. Thirdly, the prolif-erating process of a single cancer cell intrasinusoidallyor extravasating into liver parenchyma followed byproliferation (Robertson et al., 2008) under intravitalTPLSM could not be observed, as it is impossible tocontinue intravital TPLSM over 24 h due to surgicalstress. Further improvement will be needed to imagethe detailed metastatic process.

Although there are several limitations and problemsthat require resolution in our method, this is the firstreport presenting in vivo real-time images of spatio-temporal interactions between viable cancer cells andhost cells in the living liver of living mice at higher op-tical resolution and higher magnification underTPLSM.

In conclusion, high-resolution, dual-colored, real-time visualization of cancer metastasis using intravitalTLPSM can help to understand spatiotemporal tumor-

host interactions during the metastatic process in theliving organs of living animals.

REFERENCES

Erpenbeck L, Schon MP. 2010. Deadly allies: The fatal interplaybetween platelets and metastasizing cancer cells. Blood 115:3427–3436.

Gallagher DJ, Kemeny N. 2010. Metastatic colorectal cancer: Fromimproved survival to potential cure. Oncology 78:237–248.

Kedrin D, Gligorijevic B, Wyckoff J, Verkhusha VV, Condeelis J,Segall JE, van Rheenen J. 2008. Intravital imaging of metastaticbehavior through a mammary imaging window. Nat Methods5:1019–1021.

Nguyen DX, Bos PD, Massague J. 2009. Metastasis: From dissemina-tion to organ-specific colonization. Nat Rev Cancer 9:274–284.

Okabe M, Ikawa M, Kominami K, Nakanishi T, Nishimune Y. 1997.‘‘Green mice’’ as a source of ubiquitous green cells. FEBS Lett407:313–319.

Quentmeier S, Denicke S, Gericke KH. 2009. Two-color two-photonfluorescence laser scanning microscopy. J Fluoresc 19:1037–1043.

Robertson JH, Sarkar S, Yang SY, Seifalian AM, Winslet MC. 2008. Invivo models for early development of colorectal liver metastasis. IntJ Exp Pathol 89:1–12.

Starodub OT, Demitrack ES. 2008.Baumgartner HK, Montrose MH,Disruption of the Cox-1 gene slows repair of microscopic lesions inthe mouse gastric epithelium. Am J Physiol Cell Physiol 294:C223–C232.

Toiyama Y, Mizoguchi A, Okugawa Y, Koike Y, Morimoto Y, Araki T,Uchida K, Tanaka K, Nakashima H, Hibi M, Kimura K, Inoue Y,Miki C, Kusunoki M. 2010. Intravital imaging of DSS-induced cecalmucosal damage in GFP-transgenic mice using two-photon micros-copy. J Gastroenterol 45:544–553.

Wang BG, Konig K, Halbhuber KJ. 2010. Two-photon microscopy ofdeep intravital tissues and its merits in clinical research. J Microsc238:1–20.



intravital dual-colored visualization of colorectal liver metastasis in living mice using two photon...

Documents