We report a role for CXCR4 receptor trafficking in the migration of first trimester fetal MSC toward SDF-1. fMSC showed a large intracellular pool of CXCR4 protein (Figure 1), but only marginal expression of CXCR4 on the cell surface. In this basal state, internal sequestration of CXCR4 was associated with only modest migration of naïve fMSC toward SDF-1. Immunofluorescence microscopy and in situ PLA experiments examining the relationship between CXCR4 and components of the endosomal compartment showed that CXCR4 receptors were present within Rab5+ and Rab11+ endosomes, with another fraction residing within Lamp1+ late endosomes/lysosomes (Figure 2A-C) . The kinetic profile of CXCR4 trafficking in fMSC as supported by translocation modelling indicates that CXCR4 endocytosis rate regulates surface expression of CXCR4. Together this indicates a state of accelerated internalization of CXCR4 in resting fMSC (Figure 3), which when relieved by inhibition of endocytosis led to an increase in CXCR4 surface expression and potentiated fMSC migration toward an SDF-1α gradient in Transwell assays (Figure 6D).
Flow cytometry data indicate that there was up to 10-fold increased CXCR4 expression at the cell surface after blebbistatin treatment (Figure 3B and D). However this was difficult to confirm conclusively with microscopy due to the differences in sensitivity between the two techniques and the low total expression of CXCR4 in fMSC, or alternatively perhaps because inhibiting endocytosis is not the main mechanism of the increase in surface CXCR4. This 5-10-fold increase in surface CXCR4 expression, with a 2.6 fold increase in migration index, is superior to what has been reported for genetic over-expression of CXCR4. Recently Marquez-Curtis et al. reported that non-transfected human fetal umbilical cord blood-derived (UCB)-MSC had <2% CXCR4+ population . However, after transiently transfecting with a CXCR4 expression construct, 40% of cells were CXCR4+, resulting in a ~3-fold increased migration to SDF-1. Furthermore, Marquez-Curtis et al. claim their transient transfection method in fetal UCB-MSC produced superior results to other expression methods in adult rat MSC, which although increased CXCR4 expression to 54-95% of cells, but showed a mere 2–3 fold increased migration to SDF-1 [9, 15, 39].
The marked effect of blebbistatin and dynasore on cytoskeletal morphology is not unexpected as mechanistically they transiently disrupt cytoskeletal components myosin IIA and dynamin . Therefore, a small molecule with reversible effects, such as blebbistatin, is likely to be more effective at enhancing migration rather than over expression or knock down approaches to a single chemokine receptor/ligand. However, this cytoskeletal disruption also makes the cells less adherent and more sensitive to handling.
Although we only used one adult MSC donor in this study, the small number of cells expressing CXCR4 at the plasma membrane was similar to what we found with nine fetal MSC donors and as reported by others for adult MSC . Surface expression of CXCR4 reported in adult MSC has been variable, in the range of 2-25% of cells [41–43]. In keeping with this, there was a disparity between the 4% of cells we found with surface CXCR4 expression in first trimester bone marrow fMSC compared to the Jones et al. figure of 23% in blood-derived fMSC . Indeed discrepancies on levels of other fMSC markers in particular Oct4, between our collaborating laboratories was the subject of a recent review . This difference in CXCR4 expression might reflect inherent biological variation in MSC samples (donor or organ sourced, age, sampling method) coupled with differing methods of culture (serum batch) or different antibodies used in analysis.
Our finding of nuclear localized CXCR4 in bone marrow fMSC is consistent with recent findings in blood fMSC . The human CXCR4 contains a nuclear localization motif  and nuclear CXCR4 translocation has been reported to be a negative prognostic marker in several highly proliferative cancers [46–49]. Furthermore, a number of studies have found, similar to our study, that different antibody clones against CXCR4 can show disparate subcellular localization patterns, including some with nuclear localization and others with cytoplasmic localization of CXCR4 . This could be due to the epitope recognised by the antibody; for example the epitope of the CXCR4 ab2074 antibody, which we found to detect the nuclear CXCR4, is the N-terminal 20 amino acids, 9 residues of which differ between the CXCR4 mRNA splice variants 1 and 2 [51, 52]. A number of molecular weight forms of CXCR4 protein have been described by western blotting of various cells and tissues, which have been determined to be due to splice variants, dimeric receptors, and post-translational modifications [32, 53].
Fetal MSC express both CXCR4 and its ligand SDF-1 as confirmed here. Reports of an inverse correlation between CXCR4 and SDF-1 expression by MSC , is one possible explanation for our finding. That is, endogenously produced SDF-1 binds surface CXCR4, induces internalization of CXCR4, and potentially forms a suppressive autocrine loop down regulating CXCR4 expression. To investigate this we showed that treatment with a neutralizing antibody against SDF-1, resulted in only a small increase in surface expression of CXCR4 (p < 0.01). Furthermore, prolonged treatment or increased antibody concentration, which would be expected to lessen autocrine suppression, did not restore CXCR4 surface expression to a substantial level. Thus while autocrine SDF-1 may trigger some ligand-dependent CXCR4 internalization, this is not the principal mechanism responsible for intracellular localization of CXCR4 in fMSC. These data also support previous findings in non-MSC cell lines .
The chemokine receptor and other migratory mechanisms in MSC are not well understood, leading to seemingly contradictory findings in the literature [55–61]. This may be due to different migratory assays detecting different type of cellular movements. Similarly, we found that blebbistatin and dynasore did not have any effect on two dimensional migration in the scratch wound assay, but did have a significant increase in the number of cells that migrated in transwell assay in response to SDF-1. In the transwell assay, cells migrate through a membrane in response to a chemotactic ligand (e.g. SDF-1), mimicking an in vivo injury paradigm. Different migratory effects observed may also depend on the suite of chemokine receptors and ligand isoforms expressed by MSC, and the in vitro or in vivo environment . Furthermore, a number of studies ignore the capacity of CXCR4 to cross talk with other receptors directly or indirectly, especially the heterodimerizing CXCR4-CXCR7 pair [41, 62] or have alternative ligands [57, 63]. Nor do they take into account that the ligands of CXCR4 and CXCR7 homo- and heterodimeric complexes, SDF-1 and MIF, are highly expressed by MSC . Park et al. demonstrated CXCR4-overexpressing MSC displayed enhanced migration to SDF-1, but more so to glioma-conditioned media, which contains a multitude of migratory factors .