Tissue-derived mesenchymal stromal cells used as vehicles for anti-tumor therapy exert different in vivoeffects on migration capacity and tumor growth

A human cervical cancer cell line (HeLa; Cancer Research UK Cell Services, London Research Institute, Clare Hall Laboratories, Herts, UK) and human PN3 fibroblasts (kindly supplied by Dr Liu (Imperial College, London, UK)) were used. Cells were cultured in DMEM containing 10% FBS and antibiotics (Lonza, Verviers, Belgium), at 37°C in 5% CO₂.

All MSC media were supplemented with 10% FBS and antibiotics. BM-hMSCs were obtained from Lonza and maintained in DMEM low glucose (1.0 g/l) and hypoxic conditions (3% O₂). hASCs were obtained from Invitrogen (UK) and cultured in MesenPro RS Basal Medium and MesenPro RS Growth Supplement (Gibco, Paisley, UK). Human epithelial endometrium-derived stem cells or hEESCs (also known as endometrial epithelial stem cell lines; ICEp) and human stroma endometrium-derived stem cells or hESSCs (also known as endometrial stromal stem cell lines; ICEs) were supplied by Dr Carlos Simon from IVI (Valencia, Spain) [12, 13]. Cells were maintained in DMEM F-12 under hypoxic conditions (3% O₂) and dishes were pre-treated with 0.1% gelatin solution (Sigma-Aldrich Chemie GmBh, Munich, Germany). Human amniotic membrane mesenchymal stem cells or hAMCs were obtained from Cellular Engineering Technologies (CET), (Coralville, IA, USA) and were maintained in DMEM high glucose (4.5 g/l) and 10 ng/ml basic human fibroblast growth factor (hFGFb;Gibco). Cells were used between passages 5 to 8.

The hiPSCs (human IPSC line 2 F8) were kindly supplied by Dr Austin Smith (University of Cambrige, UK) and cultured in knockout DMEM (Gibco), 15% knockout serum (Gibco), 1× NEAA (Lonza), 0.1 mmol/l β-mercaptoethanol (Sigma-Aldrich), 10 ng/ml hFGFb, and antibiotics at 37°C in 5% CO₂. Cells were seeded on a PN3 feeder cell monolayer inactivated with mitomycin C (Sigma-Aldrich).

Characterization of MSCs was verified by flow cytometry. The negative surface markers used were CD45, CD34, and HLA-DR, and the positive ones were CD90, CD73, and CD105, plus CD9 and CD13 (the last two markers refers only to hEESCs and hESSCs, respectively). The antibodies used were CD13 FITC-conjugated (Immunostep, Salamanca, Spain), CD34 Percp/Cy5.5-conjugated (Becton Dickinson Co., Madrid, Spain), CD9 PE-conjugated (Millipore Corp., Billerica, MA, USA), CD45 PerCP/Cy5.5-conjugated (Becton Dickinson), CD73 PE-conjugated (BD), CD90 PE-conjugated (Becton Dickinson), CD105 FITC-conjugated (R&D Systems Inc., Minneapolis, MN, USA), and HLA-DR APC-conjugated (Immunostep). Briefly, cells were incubated in PBS supplemented with 2% FBS and specific antibodies at 4°C for 30 minutes. Then, cells were washed and fixed in 1% paraformaldehyde (Sigma-Aldrich) before FACS analysis (FACSAria system; Becton Dickinson).

For adipogenesis, MSCs were kept for 21 days in 1× basal medium (STEMPRO® Adipocyte Differentiation Basal Medium; Invitrogen Corp., Carlsbad, CA, USA), 1× supplement (STEMPRO® Adipogenesis Supplement; Invitrogen) and antibiotics. When differentiation was finished, cells were stained with Oil Red O solution (Sigma-Aldrich). For osteogenesis, MSCs were maintained for two weeks in DMEM medium containing 10% FBS, 50 μg/ml ascorbic acid, 100 nmol/l dexamethasone, and 10 mmol/l β-glycerophosphate disodium salt hydrate (Sigma-Aldrich) and antibiotics. Osteocyte formation was evaluated by staining with Alizarin Red S (Sigma-Aldrich). Images were visualized under a microscope (AE31; Motic Group Co. Ltd, Causeway Bay, Hong Kong) equipped with a camera (2500 Moticam; Motic Group) and Motic Imaging Plus 2 software (version 0.23).

The hNIS gene is endogenously expressed mainly in the thyroid and stomach, and is responsible for iodide concentration. In cells expressing hNIS, gamma ray-emitting radioisotopes such as ^99mTc are accumulated, and can be imaged by SPECT-CT, and thus the hNIS gene can be used as a reporter gene [29]. The adenoviral vector AdhNIS (also known as Ad10) used in this work was based on adenovirus serotype 5, and the hNIS gene is driven by the immediate-early cytomegalovirus promoter. AdhNIS was constructed and amplified as previously described [30]. The amount of infective adenoviral vector per cell (pFUs/cells) in culture media was expressed as multiplicity of infection (MOI). Previously, adenoviral infection efficiency was determined using adenoviral vector AdGFP testing at 100, 250, 500, and 1000 MOI, as in our previous study [31] (data not shown). For the adenoviral infection with AdhNIS, viruses were diluted in serum-free culture media to 500 MOI, added to cells, and incubated at 37°C for 1 h. The complete medium was then added and cells were maintained for 24 h until used in the in vivo experiments.

All procedures were carried out under a project license approved by the Ethics Committee for Animal Experiments from the University of Zaragoza (Spain). The care and use of animals was performed in accordance with the Spanish Policy for Animal Protection RD1201/05.

Female BALB/c nu/nu mice 6–8 weeks old (Harlan UK Ltd (Bicester, Oxfordshire, UK) and Harlan Interfauna Iberica (Barcelona, Spain)) received subcutaneous (SC) injections of 2 × 10⁶ HeLa cells suspended in 200 μl PBS for the generation of subcutaneous xenograft tumors. When these tumors reached 50 mm³ in size, mice were randomly divided into different groups, and intravenous injections of MSC were performed. For MRI experiments, animals were separated into six groups (n = 4/group). Group 1 (BM-hMSCs injected); group 2 (hASCs injected); group 3 (hAMCs injected); group 4 (hESSCs injected); group 5 (hEESCs injected); and group 6 (control; PBS injected). For SPECT-CT experiments, animals were separated into seven groups (n = 4/group), with groups 1 to 5 as above, and groups 6 and 7 being injected with hiPSCs or PBS (control), respectively.

MSCs were magnetically labeled with superparamagnetic iron oxide (SPIO; Endorem, Guerbet, France), as previously described [32]. SPIO is an oxide nanoparticle solution with a total iron content of 11.2 mg Fe/ml. Labeling with SPIO acts by reducing the transverse relaxation time on T2-weighted MRI scans. Cells were incubated with the labeling medium containing 100 μg/ml iron for 24 h. After labeling, cells were washed to remove residual contrast agent.

Animals bearing the tumor xenograft were separated into six groups (n = 4/group) when the tumors reached 50 mm³. MSCs were labeled with SPIO as described above. Groups 1 to 5 received an intravenous injection of 10⁶ SPIO-labeled MSCs, while the control group (group 6) received intravenous injection of PBS. Scans were performed at 3, 10, 17, and 24 days after injection. The MRI experiments were performed (Pharmascan system; Bruker Medical GmBH, Germany, http://​www.​bruker-biospin.​com/​pharmascan.​html) using a 7.0-T horizontal-bore superconducting magnet, equipped with a ¹H selective surface coil and a Bruker gradient insert with 90 mm inner diameter (maximum intensity 360 mT/m). All data were acquired using Paravision software (Bruker). Anesthesia was initiated using oxygen (1 l/min) containing 4% isofluorane, and maintained during the experiment with 1 to 1.5% isofluorane in O₂. T2-weighted spin-echo anatomical images were acquired by rapid acquisition with relaxation enhancement (RARE) sequence in axial (12 slices) and sagittal (8 slices) orientations and the following parameters: TR 3000 ms, TE 60 ms, RARE factor 8, average 3, FOV 30 × 30 mm, acquisition matrix 256 × 256, corresponding to an in-plane resolution of 117 × 117 μm² and slice thickness of 1.00 mm. Tumors were measured every 2 days and tumor volume was calculated using the formula:

tumor volume = 1/2 L × S², where L is long side and S is short side.

Animals bearing tumor xenograft were separated into seven groups (n = 4/group) when the tumors reached 50 mm³. MSCs were infected with AdhNIS as described above. Groups 1 to 6 received intravenous injection of 10⁶ hNIS-labeled MSCs and the control group (group 7) received intravenous injection of PBS. At 3, 10, 17, and 24 days post-injection, all groups received an intravenous dose of 18.5 MBq of ^99mTc. Anesthesia was initiated by inhaled oxygen (1 l/min) containing 4% isofluorane, and maintained during the experiment with 1 to 1.5% isofluorane in O₂. The mice were scanned using a nano-SPECT-CT scanner for small animals (Bioscan, Paris, France). A tomogram was taken, and the limits of the scan were determined. A CT and a SPECT whole-body scan were performed, with a time of 100 seconds per acquisition. The images were reconstructed with the MEDISO software (Medical Imaging Systems, Budapest, Hungary); fusion of SPECT and CT images was carried out using PMOD software (Biomedical Imagen Quantification, Basel, Switzerland). Tumors were measured every two days and tumor volume was calculated using the formula for tumor volume given above.

Quantification of radioisotope (^99mTc) accumulation was carried out using InVivoScope software (Medical Imaging Systems). Fused SPECT/CT images were used to draw the voxel-guided specific volume of interest (VOI) to accurately quantify the total activity associated with the whole tumor volume. Quantification of ^99mTc accumulation was expressed as ratio of tumor uptake to muscle uptake.

Total RNA was isolated from tumors from the in vivo SPECT-CT experiment using a commercial kit (NucleoSpin RNA II Kit; Macherey-Nagel GmBH, Dueren, Germany), in accordance with the manufacturer’s instructions. Total RNA was reverse-transcribed to cDNA (SuperScript II Reverse transcriptase; Invitrogen). Control reactions were performed by omitting the reverse transcriptase. Reverse transcription (RT)-PCR was carried out using hNIS-specific and GAPDH-specific primers (Table 1). PCR was performed using an automated system (Applied Biosystems 2720 Thermal Cycle System; Life Technologies, Glasgow, UK). cDNA products were separated by electrophoresis.

Before performing MRI scans, viability of SPIO-labeled MSCs was determined by Trypan blue exclusion assay. Cell-viability values, compared the control group (100%), with 99.23 ± 3.18% in the BM-hMSC group, 101.53 ± 5.6% in the hASC group, 96.4 ± 1.75% in the hEESC group, 100.28 ± 0.97% in the hESSC group, and 98.05 ± 6.15% in the hAMC group, thus revealing that labeling with SPIO did not significantly affect cell viability and proliferation.

The ability of MSCs to migrate in vivo towards tumor sites was first determined through MRI (Figure 2). The study of expression of surface markers by flow cytometry did not bring about changes in MSCs phenotype after SPIO labeling (Figure 3A). The recruitment of SPIO-labeled MSCs to tumors resulted in a decrease in signal intensity (SI) and visualization of darker areas in tumor sites. Tumors were visible as lighter areas. The changes in SI were detected by T2-weighted images. Although no decrease in SI was detected in control animals (data not shown), changes in SI were seen in the different MSCs injected from the first determination at day 3. Images of hASCs, hEESCs, and hESSCs at day 3 showed a higher SI decrease than hAMCs and BM-hMSCs. These differences were more pronounced from day 10, thus highlighting the differences in the ability of certain cells to migrate towards tumor areas. In all groups, SI decrease was observed first around the tumor sites and then inside the tumors, and was much more significant in hEESCs and hESSCs. To confirm these data, Prussian blue staining was used to detect SPIO-labeled MSCs in tumor sections obtained from animals scanned at day 24. Intense blue clusters were seen in tumor sections (Figure 3B). Higher-magnification images show the intra-cytoplasmic localization of the iron particles and the absence of extracellular iron. The number of Prussian blue-stained positive cells per HPF was calculated in tumor sections (Figure 3C). There were significantly more Prussian blue-stained positive cells in tumor sections of hESSCs (66.1 ± 3.72) compared with BM-hMSCs (23.6 ± 3.30).

A second in vivo imaging technique, SPECT-CT, was performed. Selected tumor transverse sections of SPECT-CT scans are shown in Figure 4A. The images obtained by SPECT-CT display a color spectrum, with the weakest signal being blue and the strongest red. In whole-body scans of control mice, an endogenous hNIS expression signal was present in the thyroid gland, salivary glands, stomach, and bladder. Accumulation of ^99mTc, reflecting hNIS expression, was detected at tumor sites in animals that received the hNIS-labeled MSC injection. No signal was detected in tumor sites of the control animals (data not shown). Because MRI assays indicated differences in signal levels at first determinations, hiPSCs were included in the SPECT-CT assays to study whether pluripotency status can affect migration ability.

hMSCs did not exert any signal at day 3, and their highest signal was found at day 10 (1.92 ± 0.72), after which it decreased until day 24 (1.07 ± 0.12). In hiPSCs and in the other hMSCs assayed, various levels of signal were detected from day 3, with hASCs presenting the lowest signal (1.44 ± 0.69) and hESSCs the highest (2.01 ± 0.18). Figure 4B shows the differences at day 3. The detected signals increased for the following determinations, with MSCs reaching their maximum level at day 17, when hESSCs also had their highest values (3.84 ± 0.41), whereas the highest signal for hiPSCs (6.73 ± 1.23) occurred at day 24. For MSCs, the detected signals decreased after 24 days, with the highest values being for hAMCs (3.07 ± 0.29) and hESSCs (2.89 ± 0.69), and the lowest for hASCs (2.46 ± 0.68) and hEESCs (2.52 ± 0.72). The tumor:muscle uptake ratio showed the differences between the migration ability of MSCs and hiPSCs (Figure 4C).

To confirm MSC and hiPSC engraftment, hNIS expression was performed by RT-PCR (Figure 5). At day 24, the animals were euthanized, and RNA was extracted from tumors as described above. hNIS expression was detected in all groups with differences of intensity. These differences may be due to the differences seen in the distribution of MSCs and hiPSCs in tumors and also to episomal hNIS expression mediated by adenovirus. Findings suggest that MSCs and hiPSCs have the ability to migrate to and engraft into tumor sites. In SPIO-labeled MSCs and in hNIS-labeled MSCs and hiPSCs, the signal was maintained for up to 24 days after injection, depending on the cell type. BM-hMSCs showed the lowest migration ability, being detected in tumor sites at day 10, whereas hiPSCs showed the highest capacity, with increased signals until day 24. The other MSC types were detected from the first determination (day 3) and the signal was stronger until the final determination (day 24), thus revealing that more cells can reach tumor areas and remain there for longer periods.

To investigate whether pluripotency may be involved in the migration ability of MSCs, analysis of the expression of pluripotency-related genes was performed (Figure 6). qPCR analysis showed that the expression of the genes studied was significantly lower in MSCs compared with the control sample, hiPSCs (taken as 100%). Moreover, differences were found in the expression levels between the MSCs. NANOG, and KLF4 expression was higher in hEESCs and hESSCs than in the other MSCs. There were no significant differences in SOX2 expression between the different MSCs, and OCT4 levels were very similar for MSCs as well, except for hAMCs in which expression was significantly lower (0.96%). REX1 was expressed only in hESSCs and BM-hMSCs (<0.01%). These results clearly show that although all cell types in the study may be considered MSCs, they exert a very different pluripotency pattern, and are also different from hiPSCs, which may be related to the differences in their engraftment capacity in addition to their effects on tumors.

When tumors reached a size of 50 mm³, approximately 10 days after HeLa injection, the animals received intravenous injection of MSCs or hiPSCs (Figure 7). Tumor size was measured until the end of the experiments (34 days after HeLa injection and 24 days after MSC or hiPSC injection). The growth of the tumors was clearly different depending on the type of MSCs or hiPSCs injected. Until day 13, the tumors of all the groups showed a similar growth pattern. However, differences between groups became clear at day 20, and were maintained until day 34. At this point, control tumors consisting of HeLa cells had an average size of 110 ± 26.8 mm³, whereas tumors composed of HeLa cells and MSCs had faster growth. Tumors resulting from the hEESC, hESSC, hASC, and hAMC injections reached a similar average size between 346 and 315 mm³, significantly larger than the control tumors except for those resulting from BM-hMSC injections (174.9 ± 17.7 mm³). Tumors composed of HeLa cells and hiPSCs had the fastest growth rates and reached an average size of 594.8 mm³, significantly larger than those in the other groups. The significantly higher growth of HeLa cells and MSC or hiPSC tumors compared with the control group indicates migration and engraftment of these cells into tumors.

To compare the expression of migration-related genes in tumors consisting only of HeLa cells (control group) and tumors consisting of HeLa cells and MSCs or hiPSCs, mRNAs encoding CXCR4, CXCL12, CCL5, CCL2, and MMP-2 were quantified by qPCR (Figure 8).

In general, expression levels of migration-related genes in tumors obtained from HeLa cells and MSC or hiPSC injection were significantly lower than those of control samples. CXCL12 expression was significantly decreased in all groups, with the smallest decrease seen in BM-hMSCs (−4.4 ± 0.04-times), and the largest in hASCs, hEESCs, and hESSCs (between −9-fold and −11-fold). For CCL5, expression levels were significantly decreased in tumors obtained from BM-hMSCs (−1.69 ± 0.25-fold), hEESCs (−2.31 ± 0.08-fold), and hESSCs (−2.58 ± 0.08-fold), but were significantly increased in hAMCs (1.73 ± 0.07-fold) and hiPSCs (0.85 ± 0.1-fold). In the hASC group, the expression level was similar to that of the control samples (−0.25 ± 0.12-fold). CCL2 and CXCR4 expression levels were significantly decreased in tumors resulting from MSC injection, with the hAMC group showing the lowest decrease (−1.05 ± 0.07-fold and −0.98 ± 0.12-fold, respectively), whereas the hiPSC group had significantly increased expression levels (5.53 ± 0.18-fold and 0.60 ± 0.33-fold, respectively). Finally, MMP-2 expression was significantly decreased in tumors composed of HeLa cells and BM-hMSCs (−6.28 ± 0.22-fold), hASCs (−6.99 ± 0.12-fold), hEESCs (−5.62 ± 0.13-fold), and hAMCs (−3.09 ± 0.06-fold) but was significantly increased in hESSCs (2.28 ± 0.07-fold) and hiPSCs (6.48 ± 0.14-fold). These findings highlight, once again, the differences between MSCs and hiPSCs. These differences may be responsible for the different migration patterns and probably play a role in the effects on the microenvironment in tumors.