LacZ transgene expression in the subcutaneous Dunn/LM8 osteosarcoma mouse model allows for the identification of micrometastasis

Authors

  • Matthias J.E. Arlt,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Ingo J. Banke,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Denise K. Walters,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Gabor J. Puskas,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Patrick Steinmann,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Roman Muff,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Walter Born,

    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    Search for more papers by this author
  • Bruno Fuchs

    Corresponding author
    1. Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland
    • Laboratory for Orthopedic Research, Department of Orthopedics, Balgrist University Hospital, University of Zürich, Zürich, Switzerland. T: + 41-44-398-1663; F: + 41-44-398-1669
    Search for more papers by this author

  • Matthias J.E. Arlt and Ingo J. Banke contributed equally to this work.

Abstract

More effective treatment of patients with metastasizing osteosarcoma (OS) with a mean 5-year survival rate of <20% requires more detailed knowledge on the complex mechanisms of metastasis for the design of new drugs, which selectively target metastasizing cells. Moreover, novel diagnostic imaging technology for early detection of metastases is needed. Mouse models, which reproduce human metastasizing OS and allow visualization of single metastatic cells are instrumental for preclinical testing of new pharmaceuticals and diagnostic instruments. Here, the low metastatic Dunn cell line and its highly metastatic LM8 subline, both equipped with a constitutively expressed lacZ gene, were used to improve the well-established OS models in syngeneic C3H mice to achieve ex vivo visualization of single metastatic cells in affected organs by X-gal staining. These models, combined with a technique for in situ high quality lung tissue-maintaining perfusion revealed, as a novel finding, single metastasizing Dunn cells in lung and liver. Importantly, constitutive lacZ gene expression did not affect in vitro and in vivo tumorigenic and metastatic properties of Dunn and LM8 cells. Thus, these improved Dunn and LM8 OS mouse models will in the future serve as a benchmark for the development of new metastasis-targeting drugs and metastasis-imaging technology. © 2011 Orthopaedic Research Society Published by Wiley Periodicals, Inc. J Orthop Res 29:938–946

Osteosarcoma (OS) is the most common primary bone tumor and the second most frequent leading cause of cancer-related death in children and young adults. At the time of diagnosis, ∼20% of OS patients have already developed detectable metastases1 while 25–50% of patients who initially present with non-metastatic disease subsequently develop metastases.2 Despite substantial improvements in surgery and chemotherapy, which have led to a dramatic increase in the survival of patients with localized disease, the 5-year survival of patients with metastases is still <20–30%.1 Although metastasis of the primary tumor to distant organs is the major cause of death in OS, the molecular mechanisms responsible for this phenomenon are still poorly understood. Moreover, the detection of metastatic cells by radiologic imaging and the success in therapeutic eradication of single metastatic cells remains limited.

Mouse models provide comprehensive insights into the complex mechanisms of tumor pathogenesis and metastasis.3 They are therefore often used in in vivo experimental cancer research and for preclinical testing of anti-neoplastic or anti-metastatic drugs. Among several OS mouse models that have been established over the last decades,4–6 the syngeneic Dunn/LM8 model is one of the most frequently used. It consists of the parental Dunn cell line, originally reported not to form metastases upon subcutaneous inoculation of the cells, and of its highly metastatic subline LM8.7 The LM8 cells were derived from the Dunn cells by eight repetitive cycles of in vivo selection for high metastatic activity according to a procedure reported by Fidler and Kripke.8 LM8 cells, unlike Dunn cells, were reported to metastasize from subcutaneous primary tumors to the lungs within 4 weeks after tumor cell inoculation with an incidence of 100%. However, numerous reports on the metastatic spread of LM8 cells are contradictory. Several groups detected metastatic foci in the lungs of mice subcutaneously injected with LM8 cells,7, 9–18 but Miyoshi et al.19 observed efficient metastasis exclusively to the liver after subcutaneous injection of LM8 cells into nude mice. In another study in nude mice, metastasis to the lung and the liver was observed after injection of LM8 cells into the tail vein.20 Moreover, Nagano et al.21 detected metastases also in the kidneys after subcutaneous and subsequent tail vein injection of LM8 cells in syngeneic C3H mice. Such variations, in part related to the use of different mouse strains and of different types and sites of tumor cell inoculation, lead to difficulties in the interpretation and comparison of therapeutic studies. Moreover, the methods used for a quantitative assessment of macro- and micrometastases are very critical but rarely reported in detail. Different and limited sensitivity of the techniques used for the detection of metastases are likely explanations for the reported differences in the metastatic dissemination of LM8 cells.

Consequently, there is great interest in standardizing and refining these methods to ultimately allow reproducible and reliable detection of single tumor cells in normal tissue. This can be achieved by equipping the tumor cells with a suitable reporter gene and by improving the quality of the host tissue for histological analysis and reporter gene detection. LacZ is a frequently used reporter gene, which encodes the bacterial enzyme β-galactosidase and is known to rarely interfere with the biological properties of the transfected cell. Cells expressing the lacZ gene metabolize the chromogenic substrate 5-bromo-4-chloro-3-indolyl-beta-D-galactopyranoside (X-gal) to an insoluble blue product. Stable constitutive expression of the lacZ gene in tumor cells represents a highly sensitive and robust histochemical reporter system for the selective recognition of the cells in normal tissue of experimental animals.22 It was reported to facilitate the monitoring of tumor progression and metastasis by enabling tumor cell detection in tissue ex vivo down to the single cell level,23–25 and it also improved the assessment of new anticancer therapies.26, 27 In addition to tumor cell reporter gene tagging, quantitative assessment of tumor metastasis to the lung in experimental mouse models can be improved by lung perfusion and fixation techniques that remove blood-related background, prevent lung collapse and maintain the morphology of functional lung alveoli. This was achieved by Borsig et al.28 who perfused the lungs of mice under anesthesia to clear them from blood and to fix and embed them in situ under inflation through the trachea.

The detection and eradication of metastatic cells in affected organs of OS patients as early as possible with novel non-invasive radiological imaging modalities and treatment strategies are of the utmost importance to improve in the future the survival of OS patients with metastatic disease. In the present study, retroviral lacZ gene transduction of tumor cells combined with in situ perfusion, fixation and embedding of lung tissue refined the Dunn/LM8 OS model and allowed for the first time the detection of micrometastases and single metastatic cells in the lung and the liver of mice. Consequently, the model will be instrumental for the future screening of new compounds for the urgently needed patient-tailored chemotherapy. Moreover, the refined Dunn/LM8 OS model will serve as benchmark for the development of novel imaging techniques with improved sensitivity and resolution for the detection of metastases in patients. Such imaging modalities for non-invasive tumor and metastases monitoring will also considerably facilitate studies with gene manipulated cells in experimental OS models designed to eventually better understand the molecular mechanisms of OS metastasis.

MATERIALS AND METHODS

Cell Lines

The mouse OS cell lines Dunn and LM8 were kindly provided by T. Ueda (Osaka National Hospital, Osaka, Japan). HEK293-T cells were obtained from the American Type Culture Collection (Manassas, VA). The cell lines were cultured in DMEM/Ham F12 (1:1) medium (Invitrogen, Carlsbad, CA) supplemented with 10% heat-inactivated FCS (cell culture medium) at 37°C in an incubator with a humidified atmosphere of 5% CO2.

Retroviral Transduction of OS Cells with a lacZ-Gene

Retroviral particles containing the lacZ gene (lacZ-retrovirus) were produced in HEK293-T cells according to an optimized protocol of Mitta et al.29 Briefly, HEK293-T cells were transfected with the following three plasmids: The retroviral expression vector pQCLIN (Clontech Laboratories Inc., Mountain View, CA), containing the lacZ and a neomycin resistance gene and the two helper plasmids pVSV-G (Clontech Laboratories Inc.), encoding the G-glycoprotein of the vesicular stomatitis virus, and pHit60 (kindly provided by Dr. Christian Buchholz, Paul-Ehrlich-Institut, Langen, Germany), coding for the retroviral gag and pol genes. Medium containing lacZ-retrovirus was collected 48 h after transfection. Dunn and LM8 OS cells were infected with lacZ-retrovirus by incubation for 48 h with virus-containing medium supplemented with 8 µg/ml Polybrene. Subsequently, the cells were grown for 10 days in tissue culture medium containing 1,200 µg/ml G418 (Invitrogen) to select for neomycin resistant lacZ gene expressing cells.

Cell Proliferation Assay

LacZ gene expressing Dunn (Dunn-lacZ) and LM8 (LM8-lacZ) cells and respective non-transduced control cells were plated in six-well tissue culture plates (12.5 × 104 cells per well). Every 24 h cells from three separate wells per cell line were trypsinized, incubated with Guava Viacount reagent and counted in a Guava EasyCyte Machine with the Viacount Acquisition Module (Guava Technologies Inc., Hayward, CA) as described.30 The experiment was repeated three times. The doubling time during logarithmic growth was calculated for each cell line.

Soft Agar Assay for Anchorage-Independent Growth

Experiments were carried out in six-well plates containing 1.5 ml bottom agar per well. The bottom agar consisted of 0.5% DNA-grade agarose (Promega, Madison, WI) containing cell culture medium supplemented with penicillin, streptomycin, amphotericin (PSA) (1:100) (Invitrogen). The plates were kept at 4°C overnight prior to use. 2 × 104 cells per well were then seeded on top of the bottom agar in 1.5 ml of 0.34% DNA-grade agarose. The plates were kept at room temperature for 5 min and subsequently transferred to a cell culture incubator. Twenty-four hours later, 2 ml of cell culture medium supplemented with PSA (1:100) were added and the cells were cultured for 21 days. Four randomly selected pictures per well, each representing a fourfold magnification of 11.2 mm2 cell culture area, were then taken with a Kappa DX20 camera (Kappa opto-electronics GmbH, Gleichen, Germany) connected to a Nikon Eclipse E600 microscope (Nikon Corporation, Tokyo, Japan). Cell colonies were counted and the size distribution was assessed with ImageJ software (http://rsb.info.nih.gov/ij/). Colony size was measured with PicedCora software (Jomessa Messsysteme GmbH, Munich, Germany).

Migration and Invasion Assay

Migration of individual OS cell lines was assessed in two-compartment 24 well plates. 2 × 104 cells in 300 µl serum-free cell culture medium were seeded in the upper compartment consisting of a filter insert with a bottom filter of 8 µm pore size (Becton Dickinson, San Jose, CA). Cell culture medium (700 µl) containing 10% FCS were added to the bottom compartment of individual wells. OS cell invasion was assessed with the same two-compartment 24 well plates with the filters in the upper compartment coated with growth factor reduced Matrigel (7.5 µg Matrigel/cm2, Becton Dickinson) according to the instructions of the manufacturer.

The number of migrating or invading cells was determined 24 h after seeding and incubation at 37°C with the following protocol: Cells attached to the upper side of the filter insert were removed with a cotton bud. Cells attached to the lower side of the insert were fixed with 10% formalin in PBS. Cell nuclei were stained at room temperature for 15 min with 300 nM Picogreen (Invitrogen) in PBS containing 50 µM digitonin. Four pictures of a filter area of 3.6 mm2 were randomly taken from each filter with an AxioCam MRm camera connected to a Zeiss AxioObserver.Z1 inverted microscope equipped with a fourfold magnification objective and an appropriate filter block for blue excitation (Carl Zeiss MicroImaging GmbH, Göttingen, Germany). Cell nuclei were counted with the ImageJ software in the four randomly selected areas and the total number of cells per filter was then calculated. Percentage migration or invasion was calculated as the total number of nuclei per filter divided by the number of seeded cells × 100. The results of individual experiments are presented as the mean of the numbers obtained from duplicate filters kept under the same experimental conditions.

Subcutaneous Mouse OS Model

Primary tumor growth and spontaneous metastasis was investigated in 8-week-old female C3H/HeNCrl (C3H) mice, which were obtained from Charles River Laboratories (Sulzfeld, Germany) at least 10 days before the begin of the experiment. Housing conditions and experimental protocols were in accordance with the guidelines of the “Schweizer Bundesamt für Veterinärwesen” and approved by the authorities of the Kanton Zurich. On day 0 of the experiment, 107 Dunn-lacZ, LM8-lacZ, Dunn, or LM8 cells were injected in 300 µl subcutaneously into the flank of the mice. The health of the mice was monitored daily and tumor length and width were measured once per week with a caliper. Tumor volume was calculated with the formula: Length × width2/2. At the end of the study 25 days after tumor cell inoculation, 50% of the mice were euthanized with CO2. The remaining mice were anesthetized and subjected to a novel protocol for in situ lung perfusion adapted from Borsig et al.28 as described in the next section.

In Situ Lung Perfusion

Mice were anesthetized with 112.5 mg/kg ketamine (Narketan®10, Vétoquinol AG, Bern, Switzerland), 16.5 mg/kg xylazine (Xylazin, Streuli Pharma AG, Uznach, Switzerland), and 15 mg/kg acepromazine (Prequilan, Fatro S.p.A., Ozzano Emilia, Italy) before the thorax and the abdomen of the mice were opened. Blood was removed from the lung and the liver by perfusion with PBS that was injected into the right ventricle of the beating heart, after sectioning the vena cava intraabdominally. When the mice had died, the lungs were fixed for 10 min with 3% paraformaldehyde (PFA) injected into the right ventricle and into the trachea with subsequent pinch-off. PFA was then replaced by injection of PBS into the right ventricle and of PolyFreeze Embedding Medium/PBS 1:1 solution into the trachea. The left lobe of the lung and the two upper and the middle lobes of the liver were processed for X-gal staining as described in the next section. The right lobe of the lung and the rest of the liver were embedded into PolyFreeze Embeeding Medium (Polysciences Inc., Warrington, PA).

Visualization of lacZ-Tagged Metastatic Tumor Cells in Lung and Liver

Organs from all mice were fixed with 2% formaldehyde and 0.2% glutaraldehyde in PBS at room temperature for 1 h, washed three times with PBS and stained at 37°C for at least 3 h in 5-bromo-4-chloro-3-indolyl-β-D-galactoside (X-gal) staining solution as described.26, 31 Pictures of whole lungs and livers were taken under a binocular microscope (OpMi-1, Zeiss, Jena, Germany) with a Kappa PS 20 C digital camera (Kappa opto-electronics GmbH) and imported as TIF files into Power Point® software. Indigo-blue stained foci with a diameter >0.1 mm on the lung and the liver of mice injected with Dunn-lacZ or LM8-lacZ cells or non-stained foci >0.1 mm on the respective organs of mice injected with non-tagged Dunn and LM8 cell were considered as macrometastases and counted. For the quantification of micrometastases, 10 randomly selected close-ups of each organ surface were taken under a microscope (Eclipse E600, Nikon Corporation) with the digital camera and imaging software described above. The pictures were then imported as TIF files into Power Point® software and indigo-blue Dunn-lacZ/LM8-lacZ foci <0.1 mm in diameter were counted in an area of 1 mm2. In organs of mice injected with control Dunn/LM8 cells, micrometastatic foci were not visible due to the lack of contrast between tumor cells and surrounding tissue.

Histology

Lungs were embedded into PolyFreeze Embeeding Medium. Cryosections of 7 µm were then obtained with a Leica CM 1850 cryostat (Leica Microsystems Nussloch GmbH, Göttingen, Germany) and mounted on SuperFrost®Plus microscope slides (Menzel GmbH & Co. KG, Braunschweig, Germany). The slides were immediately incubated in X-gal staining solution at 37°C for 24 h in a humidified chamber and then counter-stained with nuclear fast red and mounted with Immu-Mount (Thermo Electron Corporation, Pittsburgh, PA). Pictures were taken under a Zeiss AxioObserver.Z1 inverted microscope with an AxioCam MRc digital camera (Carl Zeiss MicroImaging GmbH).

Statistical Analysis

All results are presented as mean ± SEM. Data of the soft agar and the migration and invasion assays were analyzed with the Student's t-test. All other data were analyzed with the Mann–Whitney rank sum test. Data were considered significant when p < 0.05.

RESULTS

Stable expression of the lacZ gene in Dunn and LM8 cells does not affect anchorage-dependent and -independent growth.

In order to exclude effects of stable lacZ gene expression on the proliferation of low metastatic Dunn and highly metastatic LM8 cells, anchorage-dependent (Fig. 1A) and -independent (Fig. 1B) growth of Dunn-lacZ and LM8-lacZ and of the respective non-transfected (control) cells were investigated. The doubling times of adherent Dunn, Dunn-lacZ, LM8, and LM8-lacZ cell lines, calculated from growth curves shown in Fig. 1A, were 14.3, 14.0, 12.1, and 12.8 h, respectively. Thus the growth rates of lacZ gene expressing Dunn and LM8 cells were indistinguishable from the respective non-transduced cells as reflected by the corresponding congruent growth curves (Fig. 1A). However, the adherent low metastatic Dunn and Dunn-lacZ cells grew considerably slower than the highly metastatic LM8 and LM8-lacZ derivatives, resulting in significantly (p < 0.05) lower cell numbers in Dunn than in LM8 cell cultures on day 5 and 6 after cell seeding (Fig. 1A).

Figure 1.

Anchorage-dependent and -independent cell growth of lacZ-tagged and non-tagged Dunn and LM8 cells. (A) Proliferation of adherent Dunn-lacZ (▪) and LM8-lacZ cells (▴), and of non-tagged Dunn (□) and LM8 (Δ) cells. Data are mean values of triplicates ± SEM from a representative experiment carried out three times (*p < 0.05). (B) Cell growth in soft agar. The mean colony areas of Dunn-lacZ and LM8-lacZ cells and of non-tagged Dunn and LM8 cells are shown. Results are mean values ± SEM of 159 Dunn-lacZ and 238 LM8-lacZ colonies and of 81 colonies of both non-tagged Dunn and non-tagged LM8 cells of three independent experiments carried out in triplicates (*p < 0.0007).

Potential effects of stable lacZ gene expression on anchorage-independent growth of the low metastatic Dunn and the highly metastatic LM8 cells were investigated in soft agar cultures of respective lacZ gene transduced and non-transduced cells. Colony number and size were examined at 1, 2, and 3 weeks after cell seeding. At the respective time points, the total number of colonies of ≥4 cells (>20 pixels) formed by lacZ gene expressing and control Dunn and LM8 cells were indistinguishable (not shown). Moreover, after 3 weeks in culture, all four cell lines showed similar numbers of large (>2,000 pixels) colonies (not shown). However, the mean colony size of control Dunn cells was 75% larger than that of control LM8 cells (p < 0.0007) and the colonies formed by Dunn-lacZ cells were on average 59% larger than those of LM8-lacZ cells (p < 0.0001) (Fig. 1B). The cell diameters of non-adherent Dunn and LM8 cells was indistinguishable (not shown) and the increased colony size of Dunn compared to LM8 cells therefore likely due to an increased cell number per colony. Taken together, these findings point to faster anchorage-independent growth of Dunn compared to LM8 cells. This is in contrast to the results obtained with adherent cells. There, LM8 cells proliferated faster than Dunn cells. LacZ expression itself had no influence on the growth of the metastatic cell lines.

Stable expression of the lacZ gene in Dunn and LM8 cells has no effect on in vitro migration and invasion.

Migration and invasion rates were defined as the percentage of cells migrating within 24 h through non-coated or Matrigel-coated filters, respectively, of two-compartment 24 well plates. The migration rates of lacZ gene-transduced and non-transduced control Dunn and LM8 cells were indistinguishable (Fig. 2). The invasion rates of lacZ gene-expressing and control Dunn and LM8 cells were also indistinguishable, but between 61% and 69% lower than the corresponding migration rates. Taken together, the lacZ-tagged Dunn and LM8 cells had the in vitro properties of the respective non-manipulated cells and were therefore suitable for establishing a refined Dunn/LM8 OS model in syngenic mice, which allows for high sensitivity monitoring of spontaneous metastasis.

Figure 2.

Migration and invasion of lacZ-tagged and non-tagged Dunn and LM8 cells. The percentage of initially seeded cells migrating through uncoated filters (A) or invading Matrigel-coated filters (B) within 24 h was quantified as described in Materials and Methods Section. Results are the mean values ± SEM of three and five experiments carried out in duplicates with lacZ-tagged and non-tagged cells, respectively.

Primary tumor growth was not affected by constitutive lacZ gene expression but significantly faster with Dunn than with LM8 cells.

Subcutaneous injection of 107 Dunn, Dunn-lacZ, LM8 or LM8-lacZ cells per mouse resulted in the development of detectable primary tumors in all mice after 7 days. Exponential primary tumor growth was observed in all four groups of mice during the entire experimental period of 25 days (Fig. 3A). Large subcutaneous tumors without visible necrosis were observed in all four groups of mice at the end of the study (Fig. 3B).

Figure 3.

Primary tumor growth of lacZ-tagged and non-tagged Dunn and LM8 cells in C3H mice. (A) Primary tumor growth over time in mice subcutaneously injected with Dunn-lacZ (▪, n = 10), non-tagged Dunn (□, n = 10), LM8-lacZ (▴, n = 12), and non-tagged LM8 (Δ, n = 10) cells. Data at indicated time points are the mean ± SEM of calculated tumor volumes in individual mice of the different experimental groups (*p < 0.05). (B) Representative tumors in C3H mice 25 days after subcutaneous injection of indicated cell lines.

The growth curves of primary tumors derived from Dunn-lacZ and LM8-lacZ cells were comparable to those of the respective control cells (Fig. 3A). However, primary tumor growth of lacZ-tagged and control Dunn cells was considerably faster than that of tumors formed by LM8-lacZ and LM8 cells. Consequently, at the end of the experiment on day 25, Dunn-lacZ and Dunn cell-derived tumors grew to significantly (p < 0.05) larger tumors with mean calculated volumes of 1,892 ± 447 and 1,853 ± 287 mm3, respectively, compared to LM8-lacZ and LM8 cell derived tumors with respective mean volumes of 774 ± 94  and 743 ± 75 mm3.

LacZ-tagging of Dunn and LM8 cells and in situ lung perfusion and fixation enabled the detection of respective micrometastases and improved the detection of LM8 macrometastases.

As outlined previously, the original article on the Dunn/LM8 model reported that micro- and macrometastases were not recognized in C3H mice with a Dunn cell derived primary tumor.7 By contrast, the injection of LM8 cells resulted in detectable macrometastases. Here we reevaluated the Dunn/LM8 mouse OS models and reinvestigated the formation of micro- and macrometastases in lung an liver, taking advantage of Dunn-lacZ and LM8-lacZ cells and a novel protocol for in situ lung perfusion and fixation. Representative pictures of perfused and non-perfused lungs and livers of mice subcutaneously injected with lacZ-tagged and control Dunn and LM8 cells are shown in Figure 4.

Figure 4.

Detection of spontaneous lung (A, C) and liver (B, D) metastases of non-tagged (i–iv) and lacZ-tagged (v–viii) Dunn and LM8 cells in non-perfused (i, ii, v, vi) and perfused (iii, iv, vii, viii) organs in C3H mice. Metastases in representative X-gal-stained whole lungs [panels of A, C(v, vii); bar = 1 mm] and livers [panels of B, D(v, vii); bar = 5 mm] and respective close-ups [panels in A–D(vi, viii)] appear blue. Scale bars in all close-ups represent 0.1 mm. Macrometastatic foci (>0.1 mm) in the organs of mice injected with non-tagged LM8 cells [panels C, D(i–iv)] are indicated by asterisks. Areas shown in close-ups are representative for entire organs.

In mice injected with control Dunn cells, macroscopic and microscopic metastases remained undetectable in non-perfused and perfused lungs [Fig. 4A(i–iv)] and livers [Fig. 4B(i–iv)]. Interestingly, in mice injected with Dunn-lacZ cells, X-gal staining revealed blue micrometastatic foci of single cells or small cell clusters (<0.1 mm) on the surface of non-perfused lungs [Fig. 4A(vi)] and livers [Fig. 4B(vi)]. In situ perfusion and fixation of the lung and the liver further improved the detectability of Dunn-lacZ micrometastases [Fig. 4A(viii),B(viii)]. However, outgrowth to macroscopic foci was neither observed in the lung nor the liver [Fig. 4A(v, vii),B(v, vii)].

In mice injected with control LM8 cells, translucent, barely detectable macrometastatic foci larger than 1 mm in diameter were recognized in non-perfused lungs [Fig. 4C(i)] and livers [Fig. 4D(i)]. Perfusion of the lung [Fig. 4C(iii)] and the liver [Fig. 4D(iii)] did not improve the detection of the foci. However, in mice injected with LM8-lacZ cells multiple blue X-gal stained macro- [Fig. 4C(v),D(v)] and micrometastases [Fig. 4C(vi),D(vi)] were detected on the surface of non-perfused organs. Moreover, in perfused lungs, macro- and micrometastases [Fig. 4C(vii, viii)] appeared at a higher density and, as a consequence, larger number, mainly due to the translucency of the perfused tissue in which foci underneath the organ surface became visible. This effect was also observed in the liver, but it was less pronounced [Fig. 4D(vii, viii)].

The incidence of metastatic lesions in lungs and livers of all mice investigated is summarized in Table 1. In mice injected with lacZ-tagged or control Dunn cells macrometastases in lung and liver were not observed. In the group of animals injected with LM8 cells, 90% (9/10) of the mice had macrometastatic foci on the surface of the lung and 60% (6/10) on the surface of the liver. Injection of LM8-lacZ cells revealed detectable lung macrometastases in 92% (11/12) and liver macrometastases in 83% (10/12) of the mice. Micrometastases, on the other hand, remained undetectable in lungs and livers of animals injected with untagged Dunn or LM8 cells. But importantly, in mice injected with Dunn-lacZ cells micrometastases were visible in the lung of 90% (9/10) and in the liver of all (10/10) animals investigated. Moreover, all mice injected with LM8-lacZ cells had lung and liver micrometastases.

Table 1. Incidence of Detectable Macro- or Micrometastases in Lung and Liver of C3H Mice with Subcutaneous Dunn, Dunn-lacZ, LM8, and LM8-lacZ Cell-Derived Primary Tumors
Injected Cells (Mice Per Group)Mice with Detectable Metastases (%)
Macrometastases (>0.1 mm)Micrometastases (<0.1 mm)
LungLiverLungLiver
Dunn (n = 10)0000
Dunn-lacZ (n = 10)0090100
LM8 (n = 10)906000
LM8-lacZ (n = 12)9283100100

An additional histological analysis using cryosections of lung tissue confirmed the improved detectability of lung metastases in mice injected with lacZ-tagged Dunn and LM8 cells. In mice with primary tumors derived from Dunn-lacZ or LM8-lacZ cells, unlike in mice with primary tumors of the respective control cells, micrometastases or even single cell foci were recognized in lung sections (Fig. 5). Moreover, in mice injected with LM8-lacZ cells macrometastases were also more clearly visible than in animals injected with the control LM8 cells (Fig. 5C,D).

Figure 5.

Detection of Dunn and LM8 micrometastases in lung cryosections. Sections of OCT embedded lung tissue were incubated in X-gal staining solution at 37°C for 24 h in a humidified chamber and then counterstained with nuclear fast red. Arrows point to micrometastatic foci recognized in lung sections of mice injected with Dunn-lacZ (B) or LM8-lacZ cells (D). In lung sections of mice injected with non-tagged Dunn (A) or LM8 cells (C) micrometastases remained undetectable. Scale bars represent 0.1 mm.

DISCUSSION

Metastasis is the major cause of death in patients with OS. Understanding the molecular mechanisms leading to metastasis, and the capability of early detection of metastatic cells, will be key issues to improve the survival of these patients in the future. While animal models have proven to be invaluable tools to achieve these goals, the methods particularly for the analysis of metastatic spread have not been standardized and optimized to the single cell level. Inadequate evaluation of tumor development and metastatic spread in preclinical studies with animal models may lead to disappointing results in clinical trials,24, 31, 32 and it also compromises the development of improved imaging techniques.

Research focusing on OS metastasis benefits from multiple OS animal models,4–6 including the frequently used syngeneic Dunn/LM8 model. This model was used in the present study to establish ex vivo X-gal staining of metastasizing lacZ gene-transduced Dunn and LM8 cells upon in situ perfusion and fixation of the lung in the anesthetized animal, to improve the monitoring of metastasis in tissues with a main focus on the lung, the predominantly affected tissue in OS patients.

Stable transduction of tumor cell lines with a gene constitutively expressing an easily detectable reporter protein has become a common and elegant method to track metastasis in animal models. Green fluorescent protein (GFP) or red fluorescent protein (RFP) are frequently used. Luu et al.33 generated the human OS cell line TE-85 and the two derivative lines MNNG/HOS and 143B that constitutively express GFP, and injected them into the tibia of athymic mice. Regular whole body fluorescence imaging combined with X-ray imaging allowed monitoring of the primary tumor growth over time. However, lung metastases could not be visualized in vivo with this method but detected ex vivo as macroscopic foci. Recently, Kimura et al.34 have further developed and established a method for in vivo monitoring of metastasis on the surface of mouse lungs after opening the chest under anesthesia. This approach enables the monitoring of the cancer cell dynamics during metastasis in real-time. However, the procedure is technically very demanding (anesthesia, endotracheal intubation, regulation of ventilation, chest opening, re-inflation, etc.) and consequently time-consuming. In addition, fluorescent proteins have been reported to have some cytotoxic activity and to change the properties of tumor cells.35

Stable transduction of tumor cell lines with a constitutively active lacZ gene has, on the other hand and to our knowledge, not been reported to alter biological properties. This is in line with the finding in the present study that demonstrated indistinguishable proliferation in vitro and primary tumor growth in vivo of lacZ-transduced and non-transduced Dunn and LM8 cells. Thus, in vivo life fluorescence imaging as demonstrated by Kimura et al. is unlikely to become a routine method for metastasis monitoring in living animals and the here reported lacZ gene transduced Dunn and LM8 cells will be helpful in OS models complementary to those reported by Kimura et al.34

The faster primary tumor growth of tagged and non-tagged Dunn compared to the respective LM8 cells in C3H mice was inconsistent with the higher proliferation rates of adherent LM8 and LM8-lacZ cells compared to those of Dunn and Dunn- lacZ cells in vitro. Thus, the subcutaneous tissue in C3H mice, unlike the tissue culture conditions, apparently provides an (micro-)environment that supports the growth of Dunn cells more effectively than that of LM8 cells. Moreover, Dunn cells may also favor three-dimensional over adherent growth in a monolayer, as reflected by anchorage-independent growth to larger colonies in soft agar compared to LM8 cells. All findings taken together provided good evidence for no change in malignancy of Dunn and LM8 cells upon retroviral transduction with a constitutively expressed lacZ gene.

Consequently, the lacZ-tagged Dunn and LM8 cells were suitable for reinvestigating metastasis in the syngeneic mouse OS model with the aim to compare the findings with lacZ-tagged and with non-tagged cell lines. With the non-tagged Dunn and LM8 cells, the results of Asai et al.7 were largely confirmed. Both the Dunn and LM8 cells formed similar subcutaneous primary tumors, but metastatic lesions in the lung and the liver were only observed with LM8 cells. However, the mice with primary tumors of lacZ-tagged cells revealed different results. The novel finding of the present study was the identification of single metastatic Dunn-lacZ cells in the lung and the liver of mice. Moreover, in mice with LM8-lacZ primary tumors, micrometastases, in addition to macrometastases, were observed in the lungs and liver. These findings demonstrate the power of the here established X-gal staining of lacZ gene expressing OS cells combined with in situ perfusion/fixation of lung tissue in anesthetized mice for the detection of micrometastatic lesions down to the single cell level even in affected tissues other than the lung.

The novel finding that Dunn-lacZ micrometastases fail to outgrow to macrometastases remains so far unexplained, but the observations were confirmed in a separate study of experimental metastasis with intravenously injected Dunn-lacZ and LM8-lacZ cells (unpublished observation, Arlt, MJ et al.). There, Dunn-lacZ and LM8-lacZ cells were already visible on lung and liver surfaces on day 1 after injection, but seven days later only LM8-lacZ cells appeared as growing metastases, whereas Dunn lacZ cells remained silent until day 21. These and the findings of the present study therefore suggest that both Dunn-lacZ and LM8-lacZ cells metastasize to the lung, irrespective of subcutaneous or intravenous injection, however, Dunn-lacZ cells were unable to grow in the lung tissue environment. In another preliminary experiment with mice subcutaneously injected with LM8-lacZ cells, micrometastases on the surface of the lung and the liver were already visible on day 7 after cell injection and, based on the observations in the present study and the experiments with intravenously injected cells, immediately started to grow and to (macro-)metastasize. Taking all these observations together, we conclude that Dunn- and LM8-lacZ cells extravasate into the lung and the liver, but the growth of Dunn-lacZ cells, unlike that of LM8-lacZ cells, is suppressed in the lung and the liver. Thus, migration to and invasion of the lung and the liver are not the metastasis-limiting differences between the original Dunn OS cells and their highly metastatic LM8 derivatives, but rather the inability of the Dunn cells to grow in the metastatic niche. Interestingly, this is also consistent with the indistinguishable migration and invasion rates for the LM8-lacZ cells and Dunn-lacZ cells found in vitro.

In conclusion, the combination of lacZ-tagging of OS cells and in situ lung perfusion/fixation led to an enhanced detection of metastases down to the single cell level and enabled the identification of spontaneous, lung and liver micrometastases of subcutaneously injected Dunn-lacZ cells. This refined OS model will be applicable in studies aiming at the dissection of single steps of OS metastasis with gene-manipulated cells to mechanistically understand for example the process of colonization. The refined OS models will also serve as a benchmark for studies designed to improve current radiological imaging techniques such as PET and MRI to be used in clinical diagnosis and for the development of new drugs, both efforts aiming at eradicating metastatic lesions from which the majority of OS patients die.

Acknowledgements

This work was supported by grants from the Krebsliga of the Kanton Zurich, the Walter L. and Johanna Wolf Foundation, Zurich, the Lydia Hochstrasser Foundation, Zurich, the Swiss National Science Foundation, SNF, Switzerland, the Schweizerischer Verein Balgrist, and the University of Zurich.

Ancillary