Trafficking Mesenchymal Stem Cell Engraftment and Differentiation in Tumor-Bearing Mice by Bioluminescence Imaging

Abstract

The objective of the study was to track the distribution and differentiation of mesenchymal stem cells (MSCs) in tumor-bearing mice. The 4T1 murine breast cancer cells were labeled with renilla luciferase-monomeric red fluorescence protein (rLuc-mRFP) reporter gene. The MSCs labeled with firefly luciferase-enhanced green fluorescence protein (fLuc-eGFP) reporter gene (MSCs-R) were isolated from L2G85 transgenic mice that constitutively express fLuc-eGFP reporter gene. To study the tumor tropism of MSCs, we established both subcutaneous and lung metastasis models. In lung metastasis tumor mice, we injected MSCs-R intravenously either on the same day or 4 days after 4T1 tumor cell injection. In subcutaneous tumor mice, we injected MSCs-R intravenously 7 days after subcutaneous 4T1 tumor inoculation. The tumor growth was monitored by rLuc bioluminescence imaging (BLI). The fate of MSCs-R was monitored by fLuc BLI. The localization of MSCs-R in tumors was examined histologically. The osteogenic and adipogenic differentiation of MSCs-R was investigated by alizarin red S and oil red O staining, respectively. The mechanism of the dissimilar differentiation potential of MSCs-R under different tumor microenvironments was investigated. We found that the 4T1 cells were successfully labeled with rLuc-mRFP. The MSCs-R isolated from L2G85 transgenic mice constitutively express fLuc-eGFP reporter gene. When injected intravenously, MSCs-R survived, proliferated, and differentiated in tumor sites but not elsewhere. The localization of GFP⁺ MSCs-R in tumor lesions was confirmed ex vivo. In conclusion, the MSCs-R can selectively localize, survive, and proliferate in both subcutaneous tumor and lung metastasis as evidenced by noninvasive bioluminescence imaging and ex vivo validation. The MSCs-R migrated to lung tumor differentiated into osteoblasts, whereas the MSCs-R targeting subcutaneous tumor differentiated into adipocytes.

Introduction

Mesenchymal stem cells (MSCs) migrate to and proliferate within sites of inflammation and tumors as part of the tissue remodeling process. This behavior of MSCs has been exploited as a tumor-targeting strategy for cell-based cancer therapy [1–9]. However, once injected intravenously, the MSCs migrate away from the initial injection site toward tumor beds and become difficult to be visualized and tracked in vivo. Consequently, until recently the fate and migration of MSCs delivered could be identified only postmortem. Labeling MSCs with reporter gene genetically and the use of corresponding imaging modality provide a novel, noninvasive method for serially tracking and quantifying the fate of administered MSCs in vivo. The reporter gene can be expressed only in live cells and will be passed on to progeny cells. Thus, reporter gene imaging represents a powerful new approach to study the physiology and biology of transplanted cells in vivo. Examples of this approach include labeling with firefly luciferase (fLuc) or renilla luciferase (rLuc) for bioluminescence imaging (BLI) [10–12], herpes simplex virus type 1 thymidine kinase gene for positron emission tomography imaging [13, 14], and the transferrin receptor gene for magnetic resonance imaging [15, 16].

MSCs are multipotent cells that can differentiate into a variety of mesenchymal tissues, including osteoblasts and adipocytes [17]. The milieu of intracellular versus extracellular signals controls whether MSCs differentiate into osteoblasts or adipocytes. For instance, signaling through transform growth factor β (TGF-β)/bone morphogenetic protein (BMP) cytokines exemplifies extracellular mechanisms that modulate intracellular process [18, 19]. TGF-β regulates osteoblast differentiation in a biphasic manner. It stimulates development and proliferation of early osteoblasts, but inhibits the maturation and expression of phenotype-specific genes such as osteocalcin and alkaline phosphate [20–22]. In contrast, BMP2 and BMP4 cytokines are essential for osteoblasts to achieve their mature phenotype, which is characterized by the ability to form collagen-based extracellular matrix and mineral deposits. BMP2/4 cytokines positively regulate the expression of osteoblast-specific genes such as collagen and alkaline phosphatase. In addition, activation of phenotype-specific transcription factors, such as osteoblast-specific RUNX2/ Cbfa1 and adipocyte-speciifc peroxisome proliferator-activated receptor γ (PPAR-γ), determines lineage commitment [23, 24]. Adipocyte-restricted PPAR-γ transcription factor is a key regulator of osteoblast and adipocyte differentiation.

The primary purposes of this study are threefold: to investigate the tropism capacity of fLuc-enhanced green fluorescence protein (eGFP) dual reporter gene-labeled MSCs (MSCs-R) to breast cancer models; to delineate the pattern of the distribution of MSCs-R in animals with subcutaneous tumors versus lung metastases by BLI; and to examine the multilineage differentiation of MSCs-R in these two tumor models. Our data provide direct and objective evidence of the targeting and engraftment of MSCs-R to both lung metastasis lesions and subcutaneous tumor lesions when administered systemically. It is also worth noting that MSCs-R that migrate to lung-metastasis lesions tend to differentiate into osteoblasts, whereas MSCs-R that migrate to subcutaneous (s.c.) tumor lesions tend to differentiate into adipocytes. Our findings into the specific mechanisms are clinically relevant because the beneficial effects of MSCs are being tested to deliver therapy for malignancies [25, 26]; to treat osteogenesis imperfecta [27], graft-versus-host disease [28], and autoimmune diseases [29, 30]; and to improve hematopoietic engraftment [31].

Materials and Methods

Animals

Adult female Balb/C mice and male L2G85 reporter transgenic mice (Contag Laboratory, Stanford, CA) were used. Transgenic animals (L2G85) were created on the FVB background to constitutively express both firefly luciferase and enhanced green fluorescence protein (fLuc-eGFP) in all tissues and organs, including bone marrow cell populations [32]. Animal experiments were carried out according to a protocol approved by the Stanford University Institutional Animal Care and Use Committee (IACUC).

Preparation of 4Tl-rLuc-mRFP (4T1) Cells and MSC-fLuc-eGFP (MSCs-R)

The 4T1 murine breast cancer cells were transduced with a lentivirus (pFU vector) carrying a dual-fusion reporter consisting of a humanized renilla luciferase and a monomeric red fluorescent protein (rLuc-mRFP, see supporting information data) that underwent fluorescence-activated cell sorting (FACS) to establish 4T1-rLuc-mRFP (4T1) cells.

Male L2G85 reporter transgenic mice at 6–8 weeks old were sacrificed by cervical dislocation. Their femurs and tibiae were carefully cleaned from adherent soft tissue. The tip of each bone was removed with a rongeur, and the marrow was harvested by inserting a syringe needle (27-gauge) into one end of the bone and flushing with maintenance medium (Dulbecco’s modified Eagle’s medium [DMEM] with L-glutamine supplemented with 10% fetal bovine serum [FBS] and 25 µg/ml gentamicin) on poly-L-lysine (PLL)-coated flasks. The bone marrow cells were filtered through a 70-µm nylon mesh filter (BD Falcon; BD Biosciences, San Diego. http://www.bdbiosciences.com). Cells were plated into a 6-well plastic cell culture plate and cultures were kept at 37°C in a humidified atmosphere containing 95% air and 5% CO₂. The adhered cells were split when they reached 80%–90% confluence. The MSC-R surface marker expression profile was tested by FACS. The mesenchymal lineage of the MSCs-R was confirmed by their ability to differentiate into adipocytes and osteoblasts.

Flow Cytometry Analysis

MSCs-R were incubated in 2% FBS/phosphate-buffered saline (PBS) at 4°C for 30 minutes with 1 µl monoclonal antibody specific for mouse cluster of differentiation (CD)34, CD45, CD90, CD11a, CD44, and CD106 (Becton, Dickinson and Company, San Jose, CA, http://www.bd.com) or left unstained for analysis by FACSCalibur with Flow Jo software (Becton, Dickinson and Company).

Multilineage Differentiation of MSCs-R

For osteogenesis, MSCs-R were seeded at 30,000/cm² in maintenance medium on a 6-well plate and allowed to reach confluence before changing to osteogenic media (10 nM dexamethase, 25 µg/ml ascorbic acid, and 10 mM β-glycerophosphate) or fresh maintenance medium. Cells were then maintained for 21 days with a media change every 3–4 days. After 21 days, the cells were stained with alizarin red S for calcium.

For adipogenesis, MSCs-R were seeded at 30,000/cm² in maintenance medium and allowed to reach confluence before changing to adipogenic maintenance media (DMEM, 4.5 g/L glucose, 10% FCS, L-glutamine, and penicillin and streptomycin) or adipogenic media (adipogenic maintenance media with 10 µg/ml insulin, 115 µg/ml methyl-isobutylxanthine, 1 µM dexamethasone, and 20 µM indomethazine). Cells were then maintained for 21 days with a media change every 3–4 days. The adipogenic differentiation was confirmed by oil red O staining of lipid droplets.

To investigate the differentiation potential of MSCs-R in vivo, we extracted total protein from both lung metastases and subcutaneous tumors with modified RIPA lysis and extraction buffer. The total protein was quantified and diluted to the same concentration with maintenance medium. The maintenance medium containing tissues extracts was added to the MSCs-R at 90% confluence, and cultured at 37°C under 5% CO₂ for 72 hours. The total ribonucleic acid of MSCs was then extracted and analyzed for osteoblastogenic- and adipogenic-specific transcription factors, as well as for the expression of characteristic marker genes.

Study Design

Thirty-six animals were randomly divided into six groups (n = 6/ group). To study the MSC-R targeting property in lung metastasis tumor model, four groups of animals were used. In group 1, mice received intravenous (i.v.) injections of 1 × 10⁵ 4T1 cells via the lateral tail vein. In group 2, mice received i.v. injections of 5 × 10⁵ MSCs-R. In group 3, mice received coinjections of 1 × 10⁵ 4T1 cells and 5 × 10⁵ MSCs-R via tail vein. In group 4, mice received i.v. injections of 5 × 10⁵ MSCs-R 4 days after 1 × 10⁵ 4T1 cell injection. To study the MSC-R targeting property in subcutaneous (s.c.) tumor model, animals in group 5 received s.c. inoculation of 2 × 10⁶ 4T1 cells and i.v. injections of 1 × 10⁵ MSCs-R in 1 week. Subcutaneous tumor-bearing mice without MSC-R injection were used as control (group 6). When performing BLI, animals from group 1 and group 6 received intraperitoneal (i.p.) injections of d-luciferin and their BLI signals served as background. The growth of lung metastasis tumor was monitored by BLI of rLuc expression. The growth of s.c. tumor was monitored every other day with a caliper. The distribution of MSCs-R was followed by imaging fLuc expression at multiple time points after cell injection.

Bioluminescence Imaging

BLI was performed using the Xenogen In Vivo Imaging System (IVIS; Caliper Life Sciences, Hopkinton, MA, http://www.caliperls.com). The system consists of a supersensitive cooled charge-coupled device (CCD) camera mounted inside a light-tight imaging chamber. The CCD chip is 2.7 cm² consisting of 2,048 × 2,048 pixels at 13.5 µm each. The camera is capable of detecting a minimum radiance of 100 photons per second per square centimeter per steridian (p/s/cm² /sr) and can achieve a minimal image pixel resolution of 50 µm [33]. The system does not allow for three-dimensional imaging, and hence spatial resolution is limited to a compressed, two-dimensional image for analysis. Images were acquired at 1- to 10-minute intervals until the peak signal was observed. The gray scale photographic images and bioluminescence color images were superimposed using the LivingImage V 2.11 software overlay (Caliper Life Sciences). For quantification, a region of interest (ROI) was manually selected based on the signal intensity. The area of ROI was kept constant and the intensity was recorded as average photons per second per square centimeter per steridian as described previously [34].

To monitor the 4T1 tumor growth in the lung, we injected coelenterazine substrate (20 µg in 100 µL of a 10% methanol/90% PBS) via the lateral tail vein. Immediately afterward, we placed the animals supine in an IVIS system. To follow the distribution of MSCs-R, we injected d-luciferin firefly potassium salt substrate (375 mg/kg body weight in 100 µl PBS) intraperitoneally. The animals were imaged over a 20-minute time period with 1-minute acquisition intervals using IVIS system [10].

Histological Analysis

In different study groups, selected mice were sacrificed at the end of BLI session by cervical dislocation. The tissue samples were excised and weighed. Half of the tissues were fixed in 4% paraformaldehyde and embedded in paraffin for hematoxylin and eosin (H&E) staining to evaluate morphological changes. The other half of the tissues were snap-frozen and 5-µm cryosections were prepared to check for the presence of GFP⁺ MSCs-R directly under fluorescence microscope.

To detect the engraftment of MSCs-R in the s.c. 4T1 tumor, the frozen tumor sections were stained with rat anti-GFP antibody (Abcam, Cambridge, U.K., http://www.abcam.com), followed by FITC-labeled anti-rat secondary antibody. The sections were further stained with 4’,6-diamidino-2-phenylindole and observed under fluorescence microscope.

To demonstrate adipogenesis, oil red O staining was used to visualize lipid droplets as previously described [35, 36]. In brief, cultured cells or tissue samples were fixed in 4% paraformaldehyde for 30 minutes at room temperature, rinsed three times with PBS, then stained with oil red O solution (Chemicon, Temecula, CA, http://www.chemicon.com) for 50 minutes at room temperature. The sections were washed three times with water and mounted with VectaMount AQ mounting medium (Vector Laboratories, Burlingame, CA, http://www.vectorlabs.com). To demonstrate osteogenesis, mineral deposits in cultured cells and tissue samples were stained with 1% alizarin red S (Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) solution, and washed with water twice as described [37].

Statistical Analysis

Numerical data were expressed as means ± standard deviation. Linear regression analysis was performed to determine the correlation between two variables. Statistical differences between the means for the different groups were evaluated with Prism 4.0 (GraphPad Software, San Diego, http://www.graphpad.com) using either the Student’s t test or analysis of variance (ANOVA), with the level of significance at p < .05.

Results

Characterization of MSCs-R

Flow Cytometry Analysis of MSCs-R

To understand the characteristics of MSCs-R isolated from L2G85 transgenic mice, we analyzed their cell surface markers. Based on the average of three FACS analyses, MSCs were uniformly positive for extracellular matrix protein CD90, hyaluronate receptor CD44, and low-level CD106 and CD11a, but were negative for hematopoietic progenitor cell marker CD34 and leukocyte common antigen CD45 (Fig. 1A). Overall, the surface marker expression pattern of L2G85-derived MSCs-R was consistent with those reported in the literature.

Figure 1.

Characterization of mesenchymal stem cells labeled with firefly luciferase-enhanced green fluorescence protein (GFP) reporter gene (MSCs-R). (A): Fluorescence-activated cell sorting analysis showed that MSCs-R are uniformly positive for extracellular matrix protein CD90, hyaluronate receptor CD44, and low-level CD106 and CD11a, but are negative for hematopoietic progenitor cell marker CD34 and leukocyte common antigen CD45. (B): MSCs-R uniformly express GFP within the cytosol. (C): MSCs-R constitutively express firefly luciferase (fLuc). (D): Ex vivo analysis showed a linear relationship between the cell number and bioluminescence imaging signal. MSCs-R can differentiate into adipocytes and osteoblasts in vitro as assessed by oil red O staining (E) and alizarin S staining (F). Abbreviations: fLuc, firefly luciferase; RLU, relative light unit.

Linear Correlation Between Cell Number and Ex Vivo BLI Signals

MSCs-R isolated from L2G85 transgenic mice uniformly expressed eGFP within the cytosol (Fig. 1B). Representative bioluminescence images of MSCs-R are shown in Figure 1C. Ex vivo analysis showed a linear relationship between cell number and BLI signal (R² = 0.9395) (Fig. 1D). In addition, the fLuc activity did not change significantly during both adipogenic and osteogenic differentiation of MSCs-R in vitro (supporting information Fig. 1C). These data suggest that BLI of fLuc reporter gene could be used to follow and quantify transplanted stem cells in small living animals.

Multilineage Differentiation of MSCs-R

To examine the adipogenic differentiation potential of MSCs-R, cells were induced to form adipocytes (as described earlier for adipogenic potential assays) and were assessed by checking for the presence of lipid-rich vacuoles stained with oil red O As determined with this method, approximately 50% of MSCs-R had an adipocytes phenotype after 21-day adipogenic induction (Fig. 1E).

To examine the osteoblastogenic differentiation ability of MSCs-R, cells were incubated in osteoblastogenic medium. Calcium deposits were stained by alizarin red S. As shown in Figure 1F, the MSCs-R can differentiate into osteogenic cells.

Labeling the 4T1 Cells with RLuc-RFP Reporter Gene

The murine breast cancer 4T1 cells were successfully labeled with rLuc-mRFP reporter gene by lentiviral transduction. The dual-reporter gene was driven by a constitutive human ubiquitin promoter (Fig. 2A). The transduction efficiency was 73.58% based on the FACS analysis (Fig. 2B), and the RFP⁺ cells were subsequently sorted by Texas red (585 ± 25 nm) filter setting. Fluorescence microscopy of 4T1 cells demonstrated uniform mRFP expression within the cytosol (Fig. 2C). Representative bioluminescence images of 4T1 cells are shown in Figure 2D. Ex vivo analysis shows a linear relationship between cell number and BLI signal (R² = 0.90) (Fig. 2E). In addition, the mRFP expression (supporting information Fig. 1A) and rLuc activity (supporting information Fig. 1B) did not change significantly with the in vitro cell passaging.

Figure 2.

Transfection and characterization of dual-reporter gene-labeled murine breast cancer 4T1 cells. (A): Gene structure of lentivirus carrying dual-reporter (rLuc-mRFP) gene driven by a constitutive human ubiquitin promoter. (B): Lentiviral transfection efficiency was 73.58% as determined by fluorescence-activated cell sorting. (C): Fluorescence microscope image demonstrated that labeled 4T1 cells uniformly express RFP in the cytosol. (D): Transfected 4T1 cells constitutively express renilla luciferase. (E): Ex vivo analysis showed a linear relationship between the cell number and bioluminescence imaging signal. Abbreviations: LV LTR, lentivirus long term repeat; pUb, ubiquitin promoter; p/s/cm²/sr, photons per second per square centimeter per steridian; RLU, relative light unit; rLuc-mRFP, renilla luciferase-monomeric red fluorescence protein; SIN LTV, self-inactivating lentiviral long terminal repeat.

MSC-R Homing to Lung Metastasis 4T1 Tumor Model

Lung metastasis 4T1 tumor model was successfully established by i.v. injection of 1 × 10⁵ 4T1 cells as confirmed by bioluminescence imaging of the rLuc activity by IVIS system (supporting information Fig. 1). The animals were imaged at 1, 4, 7, and 10 days after 4T1 injection. The rLuc activity increased rapidly as tumor grew. We did not continue to follow the tumor growth after day 10 by rLuc BLI for two reasons: (a) the mice were very weak and could not tolerate multiple isoflurane anesthesia as a result of heavy lung tumor burden; and (b) we used the same acquisition time of 10 seconds for all image collections. Under these conditions, the rLuc signals were saturated and not suitable to quantitative analysis.

To assess MSC-R targeting and biodistribution in lung metastasis 4T1 tumor-bearing mice (supporting information Fig. 2), noninvasive imaging was performed at 1 hour after injection and on days 1, 4, 6, 9, and 11 afterward. ROIs were created over lung area, and average radiance was measured (Fig. 3). We first injected d-luciferin into 4T1 tumor-bearing mice (group 1); and no detectable optical activity was found from these mice. This result confirms that there was no cross-activity between rLuc and fLuc substrate. Therefore, we used fLuc to determine the fate of MSCs-R in vivo. In healthy mice that received MSC-R injection, bioluminescence signals associated with MSCs-R were detected primarily in lung area at 1 hour and 1 day after injection as the majority of i.v. injected cells were trapped within the pulmonary capillaries [38]. Subsequent BLI revealed that in healthy mice, the fLuc activity decreased dramatically with time (2.17 ± 0.49 × 10⁵ p/s/cm²/sr at 1 hour, 1.18 ± 0.34 × 10⁵ p/s/cm²/sr at 9 hours, and 5.28 ± 0.24 × 10⁴ p/s/cm²/sr at day 4) (group 2). In contrast, when MSCs-R were coinjected with 4T1 tumor cells, fLuc imaging signals were 1.05 ± 0.20 × 10⁶ p/s/cm² /sr at 1 hour and 9.43 ± 1.16 × 10⁵ p/s/cm² /sr at 24 hours. The BLI signal dropped to 1.27 ± 0.43 × 10⁵ p/s/cm²/sr at day 4 but further increased to 1.87 ± 0.39 × 10⁵ p/s/cm² /sr at day 6, 8.04 ± 2.15 × 10⁵ p/s/cm² /sr at day 9, and 8.37 ± 1.71 × 10⁵ p/s/cm²/sr at day 11 (group 3). The large differences in fLuc signal intensity in the lung area between tumor-bearing mice and normal mice observed 6 days after MSC-R injection are evidence of tumor-specific targeting, proliferation, and/or retention of MSCs-R.

Figure 3.

Trafficking the fate of mesenchymal stem cells labeled with fLuc-enhanced green fluorescence protein reporter gene (MSCs-R) in 4T1 tumor-bearing mice. (A): Four groups of animals were imaged to monitor the MSCs-R in lung metastasis tumor models. In group 1 (n = 6), tumor mice received i.p. injection of d-luciferin and then were subjected to BLI. The absence of BLI signal demonstrated that there was no cross-reaction between rLuc and d-luciferin. In group 2 (n = 6), healthy mice received intravenous (i.v.) injection of 5 × 10⁵ MSCs-R. The fLuc activity was detectable only in lung area up to 1 day after cell injection. In group 3 (n = 6), MSCs-R and 4T1 cells were coinjected intravenously. BLI showed that fLuc activity dropped to the lowest level at day 4, then increased gradually, and peaked at day 11. In group 4 (n = 6), mice received i.v. injection of 4T1 followed by MSCs-R in 4 days. BLI showed that fLuc activity dropped to the lowest level at 1 day after injection, and did not increase until 9 days after injection. (B): Quantitative analysis of BLI results. To monitor the homing property of MSCs-R to well-established subcutaneous (s.c.) tumors, two groups of mice were used. In group 5, s.c. injection of 4T1 cells followed by i.v. injection of MSCs-R in 1 week. The BLI images in (C) showed that fLuc activity can be detected at 6 days after injection in the s.c. tumors, then increased gradually afterward, and peaked at 14 days after injection. Quantitative analysis was shown in (D). Abbreviations: BLI, bioluminescence imaging; fLuc, firefly luciferase.

To further test the tumor targeting properties of MSCs-R, the MSCs-R were i.v. injected 4 days after 4T1 tumor cell i.v. injection. The fLuc signals were imaged and quantified to be 1.19 ± 0.45 × 10⁶ p/s/cm² /sr at 1 hour, 2.36 ± 0.13 × 10⁶ p/s/cm²/sr at 9 hours, 9.13 ± 6.99 × 10⁴ p/s/cm²/sr at day 2, 4.30 ± 0.52 × 10⁴ p/s/cm²/sr at day 4, and 4.10 ± 1.09 × 10⁴ p/s/cm²/sr at day 6. The BLI signal was then steadily increased to 7.21 ± 1.39 × 10⁴ p/s/cm²/sr at day 9 and 7.87 ± 2.06 × 10⁴ p/s/cm²/sr at day 11 (group 4). Compared with the coinjection group (group 3), fewer MSCs-R survived in the tumor and the proliferation rate of the surviving MSCs-R from group 4 is also lower than group 3.

MSC-R Targeting in the Subcutaneous 4T1 Tumor Model

Although we showed that most MSCs-R injected into healthy mice were initially trapped in the lung capillary (supporting information Fig. 3) and died within 2 days, a small number of MSCs did survive and localize to the subcutaneous tumors. Five days after MSC-R injection, the fLuc signal intensity in the subcutaneous tumor area was approximately 5.49 ± 2.71 × 10³ p/s/cm² /sr, and gradually increased afterward (1.65 ± 0.42 ×× 10⁴ p/s/cm²/sr at day 7, 2.41 ± 0.27 × 10⁴ p/s/cm²/ sr at day 8, 4.65 ± 1.52 × 10⁴ p/s/cm² /sr at day 12, and 5.04 ± 1.60 × 10⁴ p/s/cm²/sr at day 14) (group 5).

MSC-R Administration Had No Effect on the 4T1 Tumor Growth

After the completion of bioluminescence imaging, we sacrificed the mice and examined whether the injected MSCs-R had any effect on tumor growth. The mean lung weight of mice in group 1 (4T1 tumor without MSC-R injection) was significantly higher than that of mice in group 2 (healthy mice without tumor) (1.02 ± 0.06 g vs. 0.14 ± 0.004 g at 12 days after tail vein injection of 4T1 cells, p < .001). Much of this weight difference was due to the tumor tissue occupying substantial portions of the lungs of the mice injected with the tumor cells [26]. Therefore, we used whole lung weight as a surrogate endpoint of tumor burden in lungs to assess the effect of MSCs-R on the tumor growth. We found that neither the mean weight of lungs from mice coinjected with MSCs-R (group 3, 1.03 ± 0.06 g) and 4T1 cells nor that of lungs from mice injected with MSCs-R 4 days after 4T1 cell injection (group 4, 1.01 ± 0.06 g) was significantly different from that of mice injected with tumor cells only (group 1, 1.02 ± 0.06 g) (supporting information Fig. 4A). Similarly, there were no significant differences in mean tumor weights from mice injected with 4T1 only (group 6) compared with those from mice injected with MSC-R 1 week after 4T1 inoculation (group 5) in the subcutaneous tumor model (supporting information Fig. 4B). The H&E staining results consistently indicate that the injection of MSCs-R has no effect on either the lung metastasis (supporting information Fig. 4C) or s.c. 4T1 tumor morphology (supporting information Fig. 4D). These results suggest that i.v. administration of MSCs-R has no effect on the growth of 4T1 tumors both in the lungs and in subcutaneous locations in our experimental setting.

Histological Evaluation of MSC-R Homing to the Tumor

Fluorescence microscopy provided clear evidence of eGFP⁺ MSCs-R migrated to and survived in the tumor lesion and further validated our BLI measurements. In the mice that received simultaneous i.v. injections of mRFP⁺ tumor cells and eGFP⁺ MSCs-R, there were a large number of MSCs-R colocalized with the tumor cells (Fig. 4A, upper panel), whereas in the mice that received i.v. injections of 4T1 cells followed by MSCs-R in 4 days, the eGFP⁺ MSCs-R resided mainly between mRFP⁺ tumor nodules (Fig. 4A, middle panel). In contrast, we observed only minute amounts of GFP⁺ cells in healthy mice that received MSC-R injections (Fig. 4, lower panel).

Figure 4.

Localization of engrafted mesenchymal stem cells labeled with firefly luciferase-enhanced green fluorescence protein reporter gene (MSCs-R) in tumor sections. (A): Fluorescence microscope images confirmed the residence of MSCs-R in 4T1 tumor lung metastasis lesions in group 3 (upper panel) and group 4 (middle panel). In healthy mice that received MSC-R injection, only a small number of GFP⁺ MSCs-R in the lung tissue are seen (lower panel). (B): To detect the residence of MSCs-R in the s.c. tumor model, immunofluorescence staining was performed using rat anti-GFP primary antibody and FITC-anti-rat IgG as secondary antibody. In the upper panel, GFP signal was observed in sections collected from tumor-bearing mice with MSC-R injection (×100). In contrast, no GFP signal was observed in sections collected from tumor-bearing mice without MSCs-R injection (×100) (lower panel). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; GFP, green fluorescent protein; RFP, red fluorescent protein.

In the s.c. tumor model, as only a small number of MSCs migrated to and engrafted in the tumor as evidenced by BLI, it is therefore difficult to find GFP⁺ MSCs-R in tumor sections by epifluorescence microscope. To detect MSCs-R in s.c. tumor models, we used anti-GFP antibody to visualize MSCs-R. There were small patchy GFP⁺ stained areas, indicating the existence of GFP⁺ MSCs-R under low magnification (×100, Fig. 4B, upper panel). As a comparison, the 4T1 tumors from mice without MSC-R injection did not show any GFP⁺-stained cells (Fig. 4B, lower panel).

Dissimilar Differentiation Potential of MSCs-R in Lung Metastasis Versus Subcutaneous Tumor Lesions

To detect the fate of MSCs-R after homing in on tumor site, the lung tissue from groups 1–4 and s.c. tumors from groups 5 and 6 were collected and subjected to alizarin red S staining and oil red O staining to observe the osteogenic and adipogenic differentiation of engrafted MSCs-R, respectively. Lungs from animals that received both 4T1 tumor cells and MSCs-R (groups 3 and 4) were alizarin red S staining-positive (Fig. 5C), indicating calcium deposition and tissue mineralization. In contrast, the tumors taken from the mice that received MSC-R injection 1 week after s.c. tumor inoculation (group 5) were positive for oil red O staining (Fig. 5E), indicating the presence of mature adipocytes. Normal lung, 4T1-tumor bearing lung, and s.c 4T1 tumor without MSC-R injection were negative for both alizarin red S (Fig. 5D, 5G, 5H) and oil red O (Fig. 5A, 5B, 5F) staining. These findings suggest that the lung tumor metastasis niche and s.c. 4T1 tumor niche are distinct from each other and may differentially affect the commitment of MSC-R differentiation.

Figure 5.

Mesenchymal stem cell labeled with firefly luciferase-enhanced green fluorescence protein reporter gene (MSC-R) differentiation in vivo. To investigate the adipogenic differentiation of MSCs-R, we performed oil red O staining (A, B, E, F). The results showed that in the mice that received 4T1 s.c. injection followed by MSCs-R in 1 week, there were multiple red-stained lipid vacuoles in the tumor sections (E), which proved the existence of mature adipocytes. No positive oil red O staining was observed in s.c. tumors without MSC-R injection (F), or in lung tumors regardless of the MSC-R injection (A, B). To investigate the osteoblastogenic differentiation of MSCs-R, alizarin red S staining was performed (C, D, G, H). In the lung sections prepared from the mice that received intravenous 4T1 injection followed by MSC-R injection, calcium deposit (red staining) was found (C), but no calcium deposit was detected in those lung-tumor sections without MSC-R injection (D), or in s.c. tumors with (G) or without (H) MSC-R injection. Scale bar = 100 lm.

Discussion

The MSCs-R used here are characterized by their ability to proliferate in culture with homogeneous morphology, the uniform presence of a set of surface marker proteins, and their differentiation into multiple mesenchymal lineages in controlled in vitro conditions. The proper assessment of the engraftment, survival, and long-term fate of MSCs after systemic administration would require a surrogate animal model in which transplanted cells could be easily detected. We used murine breast cancer 4T1 tumor model in which the tumor cells were labeled with rLuc-mRFP and the MSCs-R expressing fLuc-eGFP. These reporter protein labels can be used for the unequivocal detection and characterization of injected MSCs-R by a variety of methodologies.

We found that MSCs-R can migrate to and survive in lung metastasis tumor. In healthy mice, the fLuc signal representing the live MSCs-R in the lungs lasted for only 1 day because the MSCs-R did not survive in normal lung. In contrast, when the MSCs-R were coinjected with tumor cells, the majority of injected cells were initially trapped in the lung. Most MSCs-R in the lung died as we observed fLuc signal decrease by ~ eightfold from day 1 to day 4. The remnant MSCs-R, however, did not passively stay in the tumor site but proliferated to some extent. The fLuc activity was increased by 1.6 ± 0.7-fold, 7.0 ± 4.0-fold, and 6.9 ± 1.7-fold at days 6, 9, and 11, respectively, compared with day 4. After 11 days of tumor inoculation, the lung tumor burden was so heavy that the mice’s respiratory function was severely impaired and they could not withstand any anesthesia. As a result, we did not further track MSC-R engraftment and proliferation. When the MSCs-R were injected 4 days after tumor cell injection, the fLuc signals were similar at 1 hour after injection, which indicates that the same numbers of MSCs-R were initially trapped in the lung capillary. However, the fLuc signals were lower than those from coinjection group at all time points afterward. In addition, the fLuc signals did not regrow until 6 days after injection. The lower fLuc signals and delayed MSC-R proliferation suggest that there were fewer MSCs-R that survived in relatively mature lung metastasis lesions, and that it took longer for this small population of MSCs to thrive. The reasons for these differences are not well understood at the present.

We also confirmed that MSCs-R are capable of targeting, engrafting, and proliferating in subcutaneous tumor lesions. We used a well-established subcutaneous 4T1 tumor model and administered MSCs-R systemically. The fLuc activity was monitored over 2 weeks. At 1 hour after MSC-R injection, the fLuc signal can be seen only in the lung area. Eight days after MSC-R injection, the fLuc signals were clearly seen in tumor site and increased gradually afterward, which indicated that small numbers of MSCs-R targeted, survived, and proliferated in s.c. tumor lesions. Note that although the scale bars used in Figure 3A and 3C are the same, the acquisition time was 1 minute for Figure 3A and Figure 5 minutes for Figure 3C. The peak fLuc signal value for group 5 was only approximately 6% and 64% of that for coinjection group (group 3) and tandem injection group (group 4), respectively.

Migration of MSCs to tumors is thought to be caused by inflammatory signaling in a tumor resembling that of an unresolved wound [39]. The innate tropism of MSCs for tumors can be exploited for the delivery of antitumor agents to the tumor microenvironment [5, 26, 40, 41]. However, the mechanism and factors responsible for the targeted tropism of MSCs to the wounded microenvironment remain to be fully elucidated. Recalling from the alternative, injury-induced cell migration models based on leukocytes and their progenitor cell line (HSC), the tumor-specific MSC migration is believed to be due to (a) inflammation-targeted homing mediated by cytokines, chemokines, and other potential chemoattractant molecules, and/or (b) hypoxia-targeted homing mediated by proangiogenic and proinflammatory molecules [42]. Our studies and reports from literature have shown that the engraftment of MSCs varies from different tumor sites and injection procedures. It is thus essential to take into consideration the location of the lesion, number of cells injected, expression level of target gene, and alternative administration route, among other factors, when designing MSC-based cancer cell therapy strategy.

We have demonstrated that systemically administrated MSCs-R have no effect on 4T1 tumor growth. Recent studies have altered the perception of the stromal cells surrounding epithelial tumors, from being innocent bystanders in the neoplastic process to being a cell type that actively promotes the growth of adjacent transformed cells. Given that MSCs participate in the formation of stroma, it is expected that transplanted MSCs would accelerate tumor growth [43, 44]. MSCs also have been reported to generate proangiogenic cytokines and enhance blood vessel growth in several studies involving ischemia [45–47]. However, coinjections of bone marrow stromal cells seem to inhibit the proliferation of Kaposi’s sarcoma cells [48]. The effect of MSCs on tumor growth is therefore equivocal at this point, varying significantly by tumor cell types, experiment settings, etc. In our experiment, we did not observe either protumor or antitumor effects of administered MSCs-R. This phenomenon might be explained by the prolific nature of 4T1 tumor cells, and the relatively short-term observation of MSC engraftment.

We observed dissimilar differentiation potential of MSCs-R in different tumor models. When the MSCs-R migrated to lung metastasis lesions, they differentiated into osteoblasts, which resulted in calcium deposit in lung tissue. On the other hand, when MSCs-R migrated to s.c. tumor, they differentiated into adipocytes, which resulted in multiple lipid vacuoles on H&E staining and positive oil red O staining. Furthermore, the tumor samples from all the other groups were negative for both alizarin red S and oil red O staining. A proper microenvironment for cellular differentiation in vivo therefore most likely involves both soluble factors as well as multiple cell-cell and cell-matrix contacts. Modulating the plasticity of MSCs and elucidating the extracellular and intracellular signaling pathways that are operating when MSCs undergo differentiation into specific cell lineages are two of the major challenges in adult stem cell research. In addition, detailed molecular dissections of the signaling pathways and transcriptional regulation determining different cellular fates would be needed. In our preliminary study, we performed preliminary semiquantitative reverse-transcription polymerase chain reaction to study the cytokine, transcription factor expression profile in different models (supporting information Fig. 5). We found that s.c. 4T1 tumor tissue expressed less BMP-2 mRNA than the cultured 4T1 cells (supporting information Fig. 5A). As BMP-2 stimulates osteoblast proliferation and inhibits adipocyte differentiation, a decrease of BMP-2 activity may lead to a decrease of MSC-derived new osteoblasts and an increase in the formation of new adipocytes. In the lung tumor model, TGFβ1 mRNA level was lower than that of normal lung and cultured 4T1 cells (supporting information Fig. 5A). Although TGFβ1 stimulates the development and proliferation of early osteoblasts, it also inhibits their maturation and expression of phenotype-specific genes. The decrease of TGFβ1 thus may result in a decrease of such inhibitory effects and lead to an enhancement of osteoblastogenic differentiation of MSCs-R. In addition, we observed that lung-tumor protein extracts can stimulate cultured MSCs-R by upregulating the expression of transcription factor RUNX2-2 (supporting information Fig. 5B), osteoblast-characteristic gene ALP, and osteocalcin (supporting information Fig. 5C). These data suggest that s.c. tumor niche and lung-tumor niche are distinct, possessing different extracellular and intracellular signaling pathways to modulate the differentiation of MSCs-R. Ongoing and future investigations of other molecules involved in the BMP/TGF and other signaling pathways will eventually help delineate the mechanism of this MSC differentiation in vivo.

There are several limitations of this study. First, we used MSCs isolated from FVB transgenic mice and transplanted them into Balb/C mice. Although there is evidence that third-party human MSC transplants are useful, there has been report in rodents that transplants cross stains would elicit immune responses [49]. This might explain in our settings the rapid clearance of MSC-based signal in the lungs and the lack of detectable signal from the liver and spleen after egress from the lung. Prior reports using both histochemistry and radiolabeling have demonstrated that murine MSCs can remain lung-resident for a few days and clearly show up in the liver/spleen thereafter [50, 51]. The different biodistribution patterns in normal animals described in the literature and tumor-bearing animals in this study encourage further investigations involving injections of FVB transgenic mice-derived MSCs to 4T1 tumors inoculated into FVB/N mice to reconcile our findings without the potential complications of immune response. Second, although the MSCs were detected in the tumors, it is likely due to two reasons: (a) the immune privileges of tumors that afford the engrafted MSCs to survive, and (b) the production of tumor-associated factors that support the MSC proliferation. There are currently no readily accepted murine MSC markers. The data presented in this study show a subpopulation of leukocyte/macrophage contamination of the MSCs. Macrophage contamination has been well documented in murine MSC isolation [52]. In fact, mouse MSCs are always contaminated with such cells unless extra steps are taken, such as a final immunoremoval of contaminating cells. It appears that mouse MSCs do not grow in the absence of CD34 cells. It is also known that macrophages home to tumors and engraft [53]. We showed that there is no colocalization of eGFP signal and F4/80+ signal (from macrophage) in both tumor samples taken either from the lung/4T1 tumor-bearing mice 11 days after MSC i.v. injection (supporting information Fig. 6) or from the s.c. tumor-bearing mice 14 days after i.v. injection of MSCs (supporting information Fig. 7), respectively. Our current data, however, do not answer whether the macro-phages contributed to the BLI signal at earlier time points. This common problem might confound our findings through bioluminescence imaging. Lastly, our data from reporter gene imaging do support the homing and subsequently proliferation and differentiation[em]whether this homing process is due to passive retention or active homing is inconclusive at the present. Although molecules such as CXCR4, CXCR12, and CCL2 have been implicated in the tissue-homing ability of MSCs [54], the exact mechanism governing MSC migration to solid tumors is still not fully characterized.

Conclusion

The mesenchymal stem cells can selectively migrate to, survive, and proliferate within both subcutaneous breast cancer and lung metastasis, in both premature and mature tumors. The subcutaneous 4T1 tumor with a low expression of BMP-2 mRNA might be responsible for adipogenic commitment of engrafted MSCs, whereas lung-tumor with low expression of TGF-β might be related to the osteogenic commitment of MSCs.