MSCs have been promoted as an attractive option for cellular delivery vehicles to carry anti-tumor agents, owing to their ability to home into tumor sites and secrete cytokines [28, 33, 34]. Previous studies have shown that systemic delivery of MSCs does not result in engraftment into healthy organs, but that they do migrate in various in vivo tumor models, although this mechanism has not yet been fully elucidated [35–42]. Additionally, MSCs can also counteract inflammation by suppressing host immune responses  and by secreting anti-inflammatory cytokines .
The ability of MSCs and hiPSCs to migrate into tumor sites after systemic injection into mice is confirmed in the present study. HeLa-based subcutaneous xenograft tumors showed extensive MSC engraftment, and BM-hMSC migration was significantly lower and slower than that obtained by the other MSCs and hiPSCs. All cell types were detected by MRI in a similar way to recent results obtained with MSCs by other groups . Signals were obtained from the first (day 3) until the final (day 24) imaging determination, except in BM-hMSCs, which highlights the capacity of these cells to persist in tumor sites. These results and the differences between BM-hMSCs and the other cell types were confirmed by iron-particle staining and the detection of hNIS in tumor samples. Using SPECT-CT images and hNIS expression, another study has shown BM-hMSC migration towards a breast-cancer model . In contrast to our observations, hNIS expression in that study was detected in tumors from day 3 after MSC injection, which may be attributed to the differences between donors, tumor models, and the size of the pre-established tumor.
To clarify whether pluripotency plays a role in MSC migration to tumors, a pluripotency marker panel was analyzed. These studies revealed diminished expression of SOX2, NANOG, OCT4, KLF4, and REX1 compared with the control sample (hiPSCs). hEESCs and hESSCs, which displayed the highest signals in the imaging techniques, also had significant expression of KLF4, NANOG, and REX1 with respect to the other MSCs, although it was much lower than that shown by hiPSCs. These differences in pluripotency patterns may influence the differences in migration and engraftment of MSCs and hiPSCs.
Tumor growth confirmed the ability of MSCs to migrate into tumors, and the significant differences between BM-hMSCs, the other MSCs, and hiPSCs. In a previous PC3 prostate xenograft model , intra-tumoral injection of hASCs also induced larger tumors compared with the control group, although the differences between control and hASC-induced tumors were smaller at the final time point (<2-fold vs. 3-fold in ours). These differences may be attributed to the different tumor model used and the different procedures for hASC injection. However, there are also studies reporting inhibitory effects on tumor growth in an in vivo tumor model following MSC injection, using different approaches from ours [46, 47]. The dose of MSCs delivered and the timing of injection have been highlighted as determining factors in the promotion or inhibition of tumor growth .
Although the mechanisms behind homing are not yet fully understood, the most likely cause of preferential migration is the release of chemotactic gradients from tumors. MSCs have a wide range of chemokine and cytokine receptors on their cell surface, which respond functionally to their ligands in vitro, whereas in vivo, their modification implies changes in migration behavior [48–52].
Some authors have shwon that MSCs secrete a large panel of chemokines such as CXCL12, CCL2, and CCL5, which implies activation of the MAPK, FAK, and STAT signaling pathways, and the induction of biological responses [48, 49]. Moreover, tumors produce a wide range of chemokines and cytokines, which may act as ligands for MSC receptors [35, 53]. Of the different pairs of receptor/ligand described as responsible for migration, the CXCL12/CXCR4 pair should be highlighted. This has been studied in numerous works, both in vitro and in vivo, and using different tumor models or MSCs sources [35, 49–51, 54–59]. Loss of expression in MSC´s receptors  implies a loss in cell-migration ability, indicating the great importance of these axes in MSC migration.
The relationship between homing and the inflammatory state has been assessed, and studies have been conducted in which BM-hMSCs were pre-treated with factors involved on inflammation (such as tumor necrosis factor, MMP2, CXCL12 and CCL5) . But some studies were contradictory [35, 50], which it could be due to the variation in donors or cell-culture conditions (confluence, hypoxia, and passages) .
Our results show that, in general, when MSCs engraft into tumors, migration-related gene expression is decreased (whereas hiPSCs exert an increase) except for the CXCL12/CXCR4 axis, causing a significant depletion in the expression of both markers when MSCs are present. The exceptions were hiPSCs, in which CXCR4 expression increased. The reason for this may be that CXCL12 expression is also linked to more immature cell fractions, with higher expression in less committed stages of differentiation, that is, in cells closer to the embryonic state . With regard to hiPSCs, this study and another previous work  are the first to highlight their in vivo migration ability to and long-term engraftment into tissue damage areas.
Besides the release of chemotactic gradients from the tumors, other explanations for MSCs migration could be the hypoxic conditions produced by tumor cells, which may cause MSCs to increase the expression of migratory signals , and thus confer the ability to cross the biological barriers . However, in our study, a direct correlation between the migration patterns and the absence of oxygen does not seem to exist (data not shown).
Finally, a recent dialogue mediated by exosomes between MSCs from the bone marrow and tumoral cells in patients with melanoma  has been shown. The degree of exosome release by the various MSCs could determine the migratory differences between those cells, depending both on their area of origin and also on their exosome targeting.